0 INFO-VAX	Sun, 27 Jan 2002	Volume 2002 : Issue 52      Contents:D Re: Capellas redefines Industry Standard to mean Windows *and* LinuxD Re: Capellas redefines Industry Standard to mean Windows *and* LinuxD Re: Capellas redefines Industry Standard to mean Windows *and* Linux Re: Compaq financial conference  Re: Compaq financial conference  Re: Compaq financial conference  Re: Compaq financial conference  copy subdirectory  Re: copy subdirectory  Re: copy subdirectory  Re: copy subdirectory  Re: copy subdirectory 3 Re: DECdtm DTI$S_PART_NAME should be (no *IS*) 255!  http://birthdaybill.81832.com ! Re: I/O Performance on Alpha 4100 ! Re: I/O Performance on Alpha 4100  RE: Kerberos in VMS  Re: Makefile Re: Makefile Re: mount problems on cluster  Re: mount problems on cluster  Re: mount problems on cluster ( RE: new partner also sent to openvms.org( Re: Northernlight to close public access! passing /debug to linker from MMS % Re: passing /debug to linker from MMS % Re: passing /debug to linker from MMS  Re: Selling VMS licensesP Re: Sv: Sv: Sv: Younger recruits versus experienced veterans ( was Re:     The d TCP/IP problems  Re: TCP/IP problems  Re: TCP/IP problems  Re: TCP/IP problems  Re: TCP/IP problems  Re: TCP/IP problems 3 Re: VMS system on the sourceforge.net Compile Farm? B Re: Younger recruits versus experienced veterans ( was Re: The dem  F ----------------------------------------------------------------------  % Date: Sun, 27 Jan 2002 19:16:01 +0010 % From: paddy.o'brien@zzz.tg.nsw.gov.au M Subject: Re: Capellas redefines Industry Standard to mean Windows *and* Linux 5 Message-ID: <01KDKRUS0YOI002PIW@tgmail.tg.nsw.gov.au>    Alan Greig said:  B >In the financial conference Capellas has just said (I paraphrase)A >"there is absolutely no question that industry standard systems   >(slide @ >defines this as Windows and Linux) are eviscerating proprietary1 >systems. The era of proprietary systems is over"   D I trust "eviscerating" was his word not part of your paraphrasing.  ? Since it seems to make no sense -- what does "disembowelling a  @ proprietary system" mean? -- I can understand such ignorant and A pretentious vocabulary after reading some of his other "piece de   majeste", but not from you.   ! >Better open source VMS now then.  > < >Interestingly Capellas seems to be defining Windows as non  >proprietary& >again. Someone should tell Microsoft.  D But didn't King Billie III, only a short while ago, claim that open E source would not have existed without Microsoft?  I could never work  % out what he intended to mean by that.    Regards, Paddy   ------------------------------    Date: 27 Jan 2002 06:18:33 -0800( From: bob@instantwhip.com (Bob Ceculski)M Subject: Re: Capellas redefines Industry Standard to mean Windows *and* Linux = Message-ID: <d7791aa1.0201270618.3694875e@posting.google.com>   Y JF Mezei <jfmezei@videotron.ca> wrote in message news:<3C535BE7.E601C793@videotron.ca>...  > Bob Ceculski wrote: O > > not too expensive for the little guy to run vms ... do you think windows is O > > cheaper after buying 80000 boxes per one alpha vms server and then spending & > > 80% of your time in patching bugs! > O > The problem for small shops is that due to lack of available software on VMS, J > they will also need to maintain another architecture to run the apps not > available on VMS.  > B > If you have a critical app that is big enough to warrant its ownH > infrastructure, then VMS can be justified., But if you have many smallN > applications, none of which "requiring" VMS and many which don't run on VMS,U > it becomes harder to justify VMS unless you can run both big and small apps on VMS.  > M > And because VMS has no public visibility, one needds to work much harder to L > sell a VMS solution to a potential smaller customer who has no idea of theN > value of VMS. And the fact that Compaq doesn't publicly sell VMS, it is also@ > hard to spin the situation positively in terms of VMS' future. > L > We need Compaq's help to penetrate new markets. And right now, not only isM > Compaq not providing the help, but it is working against those who wants to M > push VMS solutions in markets other than the few remaining ones that Compaq  > tolerates VMS in.   K the size of the app should never matter ... security and relaibility should C always be a priority when building any system, no matter what size!    ------------------------------  # Date: Sun, 27 Jan 2002 18:21:32 GMT * From: "Bill Todd" <billtodd@metrocast.net>M Subject: Re: Capellas redefines Industry Standard to mean Windows *and* Linux C Message-ID: <MyX48.144525$QB1.10723761@bin3.nnrp.aus1.giganews.com>   2 <paddy.o'brien@zzz.tg.nsw.gov.au> wrote in message/ news:01KDKRUS0YOI002PIW@tgmail.tg.nsw.gov.au...  > Alan Greig said: > D > >In the financial conference Capellas has just said (I paraphrase)B > >"there is absolutely no question that industry standard systems	 > >(slide B > >defines this as Windows and Linux) are eviscerating proprietary3 > >systems. The era of proprietary systems is over"  > D > I trust "eviscerating" was his word not part of your paraphrasing.@ > Since it seems to make no sense -- what does "disembowelling aA > proprietary system" mean? -- I can understand such ignorant and B > pretentious vocabulary after reading some of his other "piece de > majeste", but not from you.   H Perhaps Curly's been taking lessons from Mike "soft underbelly" Winkler.   - bill   ------------------------------  # Date: Sun, 27 Jan 2002 13:47:52 GMT ? From: Jim.Johnson@software-exploration.nospam.com (Jim Johnson) ( Subject: Re: Compaq financial conference/ Message-ID: <3c5400a4.8381862@news.demon.co.uk>   C I've been sitting here fascinated by this thread, and not sure if I  want to jump in or not.    First, the short answers:   - - Bill is absolutely correct.  Listen to him. F - Read the basic papers and books behind these technologies, if you'reC going to be working in, or selling in, that space.  Don't trust the E marketing droids or the brochures any more about this technology than 6 you would about, say, OS scalability and robustness...  $ Second, the slightly longer answers:  A - As Bill pointed out, there are a number of completely different  issues here:  E -- Transactions: These are *defined* as having ACID properties, which > include durability of the result.  Anything that doesn't offer6 durability of the result isn't offering a transaction.  ? -- Availibility: This is a function of MTBF and MTTR (Mean Time C Between Failure, and Mean Time To Recover, respectively).  Clusters ? offer higher availibility than single node systems by radically E reducing MTTR, although at some cost to MTBF (ok, in V4.0 it was at a A significant cost to MTBF).  High availibility does not imply that ? every operation will succeed, or even that *any* operation will > succeed.  It simply says that you'll have a high likelihood ofA submitting an operation.  Commercially, a viable HA system should ; suggest that such operations will have a high likelihood of 6 completing, but that is a completely different matter.  D -- Fault Tolerance: This specifies that there are an increased classB of failures that will be silently absorbed and accounted for.  TheD class of failures may be hardware-biased (e.g. the ftVAX), or may beE software-biased (e.g. the shuttle with its one independently designed D and coded control system variant voting on the result -- or at leastA that's how it worked in the 80's...  I haven't checked recently). ? Note that it offers no guarentees outside of the failures it is > designed to detect and handle (has anyone heard about 'nonstopA failures' that you can get on a Tandem due to process pairing and  software faults?  ' Regardless, Fault Tolerance depends on:   F -- Fail Fast:  This says that failures will be detected 'immediately',D for some value of immediately, and the system halted.  Without that,B you can have a system get a failure, corrupt the system state, and, stumble on for some distance before failing.    F Now, Bill in a reply suggested reading Gray & Reuter.  I fully concur. It is an excellent book.  E I would also add, if you can get ahold of it, a small paper published A in 1981 that defined transactions.  It's "Atomic Transactions" by @ Butler Lampson.  It was published by Springer Verlag in a volume5 entitled "Distributed Systems -- An Advanced Course".   E It's a wonderful paper, inasmuch as it spends its time describing the B semantics of the software structures necessary to get from systemsE that fail to transactions.  I will warn you that it does not describe F a commercially viable solution, but that is not its point.  It's stillC the best paper I've ever read on what transactions do and don't do.    Jim.        F On Sat, 26 Jan 2002 00:53:36 GMT, "Bill Todd" <billtodd@metrocast.net> wrote:   > 7 >"Main, Kerry" <Kerry.Main@Compaq.com> wrote in message M >news:BE56C50EA024184DAF48F0B9A47F5CF401AB1BEA@kaoexc01.americas.cpqcorp.net.  >..  > I >>>> If it's 'purist' to use terms only according to their definition, so  >be it<< > G >Well, it depends if one views it from a hardware / OS perspective or a / >higher level business transaction perspective.  > F >Different approaches to the same objective i.e. how does one ensure a/ >transaction is never lost once it is commited?  >  >***@ >No, they are not different approaches to the same objective.  AL >fault-tolerant system ensures that system operation will survive a hardwareI >fault without producing incorrect results, and by itself says absolutely G >nothing about the ability of a committed transaction to survive a site J >disaster (which is accomplished through high-availability mechanisms thatH >are orthogonal to fault-tolerance, such as remotely-mirrored storage or
 >systems). > H >A transaction *by definition* is never lost once it has been committed,K >unless the storage on which it has been made persistent is lost.  That has L >nothing to do with fault-tolerance *or* availability:  it is the definitionK >of how transactions work.  Increased availability means that the result of C >the committed transaction will have a greater probability of being G >accessible, and fault-tolerance means that the result of the committed G >transaction will have a greater probability of being correct (and also J >increases the probability that a local hardware failure will not make theE >result inaccessible - but this has nothing to do with surviving site  >disasters). >  >- bill  >  >  >    Jim Johnson  Software Exploration, Ltd.) (remove '.nospam' from the reply address)    ------------------------------  % Date: Sun, 27 Jan 2002 11:12:30 -0500 % From: JF Mezei <jfmezei@videotron.ca> ( Subject: Re: Compaq financial conference, Message-ID: <3C5426E9.E9D49779@videotron.ca>   Jim Johnson wrote:G > -- Transactions: These are *defined* as having ACID properties, which @ > include durability of the result.  Anything that doesn't offer8 > durability of the result isn't offering a transaction.  L Here is an example of a "failure" I personally experienced in the early days) of interbank ATMs back in 1986 timeframe.   N My account is with Bank A. I go to Bank-B's ATM to make $60 widthdrawal. AfterL I press "OK", machine waits for a while, screen goes blank, card is ejected,K and then screen goes back to the "welcome" screen.  So I do the transaction C again, with the same result. No receipt was printed in either case.   K Later, I called the my bank (that was before the 24 hour banking days, so I H called the "lost card" dept" to advise them of the situation and low andE behold, they confirm that I succesfully widthdrew $60 twice in a row.   N It was only because my job did put me in contact with a security manager at myK bank that I got my money back because my bank had not gotten the "money was K NOT handed out" message from Bank-B due to their computer failures and from F their pont of view, they had received proof of transaction completion.  K I ended up working for bank-B later on and was good buddies with the tandem L folks who told me that the types of problem I had experienced was one of theL reasons that they had been able to justify getting the tandem infrastructureG against the wishes of the folks who wanted nothing but IBM at the bank.     F > -- Fault Tolerance: This specifies that there are an increased class? > of failures that will be silently absorbed and accounted for.   I No, it specifies that the software will handle failure or corruption of a M process and keep a transaction which is in progress in a known state that can  be recovered/continued.   M Tandem not only provides redundant hardware with hardwrae controllers to hide K many hardware failures, but it also provides an OS that has both management M and software *TOOLS* to allow fault tolerance.  If you login interactively on N the tandem and the processor fails, your editor session will fail, it will not, continue transparently on another processor.  N The fault tolerance is provided at the application layer. The application MUSTI make use of the tools provided by the OS to allow itself to have multiple J cooperating instances where one instance can CONTINUE the work if a failedG instance. Note than an instance can fail not only for hardware but also  software reasons.   I Note that shared memory in a Wildfire between VMS instances theoretically K allows software to become fault tolerant and a backup process to continue a  transaction from another node.  G > software-biased (e.g. the shuttle with its one independently designed F > and coded control system variant voting on the result -- or at least' > that's how it worked in the 80's...     L Nop. There is no software voting on the shuttle. The "voting" is done at theM muscle level. If you have 3 hydraulic pistons to move a control surface, each I is controlled by a different computer. If one computer fails or sends the L wrong command to the piston, the 2 other computers will out-muscle the first one.    L The 5th computer does have different software, but is only engaged manually.N Otherwise, all it does is keep situation awareness by listening to the trafficK on the various buses.  If one of the 4 main computers fails, it takes a few H minutes to manually "re-string" the bus to distribute the device control authorities to the other 3.   N And there is a very good reason for this. You do not want a single computer toI end up controlling 2 of 3 hydraulic pistons that move a device. (At which + point it can outmuscle the other computer).   I The computers are not fault tolerant. They do not share memory. They just M listen in promiscuous mode to the traffic on the bus. But computer B does not M have authority to control a device that was assigned to computer A unless the . bus is restrung to give that authority to "B".   ------------------------------  # Date: Sun, 27 Jan 2002 18:30:02 GMT ? From: Jim.Johnson@software-exploration.nospam.com (Jim Johnson) ( Subject: Re: Compaq financial conference0 Message-ID: <3c544136.24911861@news.demon.co.uk>  C On Sun, 27 Jan 2002 11:12:30 -0500, JF Mezei <jfmezei@videotron.ca>  wrote:   >Jim Johnson wrote: H >> -- Transactions: These are *defined* as having ACID properties, whichA >> include durability of the result.  Anything that doesn't offer 9 >> durability of the result isn't offering a transaction.  > M >Here is an example of a "failure" I personally experienced in the early days * >of interbank ATMs back in 1986 timeframe. > O >My account is with Bank A. I go to Bank-B's ATM to make $60 widthdrawal. After M >I press "OK", machine waits for a while, screen goes blank, card is ejected, L >and then screen goes back to the "welcome" screen.  So I do the transactionD >again, with the same result. No receipt was printed in either case. > L >Later, I called the my bank (that was before the 24 hour banking days, so II >called the "lost card" dept" to advise them of the situation and low and F >behold, they confirm that I succesfully widthdrew $60 twice in a row. > O >It was only because my job did put me in contact with a security manager at my L >bank that I got my money back because my bank had not gotten the "money wasL >NOT handed out" message from Bank-B due to their computer failures and fromG >their pont of view, they had received proof of transaction completion.: >eL >I ended up working for bank-B later on and was good buddies with the tandemM >folks who told me that the types of problem I had experienced was one of thenM >reasons that they had been able to justify getting the tandem infrastructurecH >against the wishes of the folks who wanted nothing but IBM at the bank. >o   Two points here:  F First, the brutal truth around ATM machines is that the machine itselfB is engaged in a non-recoverable activity (handing over money), andC therefore that activity must occur without transactional guarenteese? (specifically you can't guarentee atomicity or consistency in a0D sequence of database operations and the ATM real world action).  You? need something with fault tolerance,like Tandems, and very wells> designed ATM terminals, to reduce the chances of what you have
 described.  E Second, depending on what 'computer failure' they were talking about,bE it also could be that the interbank activity was operating outside oflD a transaction, or a sequence of escrowable transactions.  That wouldC be possible, but ... well ... highly questionable.  I'm assuming it  was the first.   > G >> -- Fault Tolerance: This specifies that there are an increased class @ >> of failures that will be silently absorbed and accounted for. >rJ >No, it specifies that the software will handle failure or corruption of aN >process and keep a transaction which is in progress in a known state that can >be recovered/continued. >g  D It may mean that for some cases in some instances.  It may not.  ForE instance, let's assume I am on a platform that provides pure hardware.@ fault tolerance.  That platform will identify and compensate forE processor and memory hardware failures, thus allowing the software tod continue correctly.s  F However, on that same platform the arrival of your transaction induces= a timing related Heisenbug that causes my process to throw anrA exception.  Nothing on that platform will catch and automaticallyi> correct that failure.  That transaction will fail partway (and@ therefore get rolled back, since it is a transaction -- this hasB nothing to do with the platform being fault tolerant or not) and aE failure returned to the invoking agent.  In turn, that invoking agent E _may_ have a policy of resumitting for some classes of errors or someaD number of times.  That has nothing to do with either fault tolerance or transactions.  E However, note that nothing about that fault tolerant platform handledf that class of failures.   @ So, the class of failures that could be handled and absorbed wasB increased.  It was not increased to 100% of all possible failures. Yet it is still fault tolerant.   E Transactions, fault tolerance, and techniques such as process pairingiF are all good things, and they may have mutual benefits.  However, they! are not inherently tied together.t  N >Tandem not only provides redundant hardware with hardwrae controllers to hideL >many hardware failures, but it also provides an OS that has both managementN >and software *TOOLS* to allow fault tolerance.  If you login interactively onO >the tandem and the processor fails, your editor session will fail, it will nott- >continue transparently on another processor.  >hO >The fault tolerance is provided at the application layer. The application MUSTlJ >make use of the tools provided by the OS to allow itself to have multipleK >cooperating instances where one instance can CONTINUE the work if a failedaH >instance. Note than an instance can fail not only for hardware but also >software reasons. >hJ >Note that shared memory in a Wildfire between VMS instances theoreticallyL >allows software to become fault tolerant and a backup process to continue a >transaction from another node.t >i  F Oh, I agree that the Tandem tools are incredibly useful.  Don't get meF wrong.  For quite a while I've thought the lack of their equivalent on  a VMS system was a real problem.  E My only point is that they're just simply not part of a definition of  a transaction.    H >> software-biased (e.g. the shuttle with its one independently designedG >> and coded control system variant voting on the result -- or at leastt( >> that's how it worked in the 80's...   > M >Nop. There is no software voting on the shuttle. The "voting" is done at therN >muscle level. If you have 3 hydraulic pistons to move a control surface, eachJ >is controlled by a different computer. If one computer fails or sends theM >wrong command to the piston, the 2 other computers will out-muscle the firstd >one.  >iM >The 5th computer does have different software, but is only engaged manually.pO >Otherwise, all it does is keep situation awareness by listening to the traffictL >on the various buses.  If one of the 4 main computers fails, it takes a fewI >minutes to manually "re-string" the bus to distribute the device controli >authorities to the other 3. >dO >And there is a very good reason for this. You do not want a single computer tosJ >end up controlling 2 of 3 hydraulic pistons that move a device. (At which, >point it can outmuscle the other computer). > J >The computers are not fault tolerant. They do not share memory. They justN >listen in promiscuous mode to the traffic on the bus. But computer B does notN >have authority to control a device that was assigned to computer A unless the/ >bus is restrung to give that authority to "B".b  B This one I'll give you.  It was over 20 years ago when I last readA about the shuttle.  I'll assume I remembered that one wrong.  Oldn neurons.   Jim Johnson  Software Exploration, Ltd.) (remove '.nospam' from the reply address)6   ------------------------------  # Date: Sun, 27 Jan 2002 18:19:22 GMT * From: "Bill Todd" <billtodd@metrocast.net>( Subject: Re: Compaq financial conferenceA Message-ID: <JwX48.59926$vH6.3089752@bin6.nnrp.aus1.giganews.com>   2 "JF Mezei" <jfmezei@videotron.ca> wrote in message& news:3C5426E9.E9D49779@videotron.ca... > Jim Johnson wrote:I > > -- Transactions: These are *defined* as having ACID properties, whichoB > > include durability of the result.  Anything that doesn't offer: > > durability of the result isn't offering a transaction. >eI > Here is an example of a "failure" I personally experienced in the earlyi days+ > of interbank ATMs back in 1986 timeframe.a >iJ > My account is with Bank A. I go to Bank-B's ATM to make $60 widthdrawal. AfterrE > I press "OK", machine waits for a while, screen goes blank, card isT ejected,A > and then screen goes back to the "welcome" screen.  So I do theu transaction E > again, with the same result. No receipt was printed in either case.T >aK > Later, I called the my bank (that was before the 24 hour banking days, soT I.J > called the "lost card" dept" to advise them of the situation and low andG > behold, they confirm that I succesfully widthdrew $60 twice in a row.  >iJ > It was only because my job did put me in contact with a security manager at my I > bank that I got my money back because my bank had not gotten the "money  washH > NOT handed out" message from Bank-B due to their computer failures and fromH > their pont of view, they had received proof of transaction completion. >aF > I ended up working for bank-B later on and was good buddies with the tandemJ > folks who told me that the types of problem I had experienced was one of the ? > reasons that they had been able to justify getting the tandema infrastructureI > against the wishes of the folks who wanted nothing but IBM at the bank.   J That's a very good example of the value of transactions:  the behavior youG experienced could *not* have occurred had it actually been handled as aeK correctly-distributed transaction using the standard industry definition ofeL the term (which is a *lot* more stringent than the common English definition
 of the term).v  L In particular, when a physical action (such as dispensing cash) is involved,K if you use transactions correctly the *only* case in which there can be anybC question about whether or not that action occurred is when the nodetH *performing* that action fails (and that node should be designed - e.g.,K using a small amount of redundant flash memory - such that when it recoversiL it can correctly ascertain what it did).  Otherwise, the node that performedH the action always knows what it did, and the other nodes involved in theG transaction never assume anything (if they fail, when they recover theyn0 ascertain what happened from someone who knows).   >t >eH > > -- Fault Tolerance: This specifies that there are an increased classA > > of failures that will be silently absorbed and accounted for.o >.K > No, it specifies that the software will handle failure or corruption of a K > process and keep a transaction which is in progress in a known state that@ canc > be recovered/continued.a  G As I said, you don't understand the definition.  Typical fault-tolerantDG systems have no software component whatsoever:  they rely upon hardwareIJ mechanisms (pair and spare take-over or TMR voting) to make the underlyingL hardware appear perfect to the software that runs on it.  This does have theJ effect of keeping that software running without perceptible pause, but hasD nothing to do with transactions (which aren't recovered or otherwise> affected:  they just continue to run like any other software).   >nJ > Tandem not only provides redundant hardware with hardwrae controllers to hideB > many hardware failures, but it also provides an OS that has both
 managementL > and software *TOOLS* to allow fault tolerance.  If you login interactively onL > the tandem and the processor fails, your editor session will fail, it will notm. > continue transparently on another processor. > ; > The fault tolerance is provided at the application layer.s   No, it is not.  See below.    The application MUSToK > make use of the tools provided by the OS to allow itself to have multipleTL > cooperating instances where one instance can CONTINUE the work if a failedI > instance. Note than an instance can fail not only for hardware but also  > software reasons.i  I Yes, it can - though if it fails for software reasons, there's at least afI good chance that its partner will fail for those same reasons (that's whya: systems that actually claim to address software faults useI independently-developed software in each instance - and since Tandem doesl= not, it's not technically more than hardware-fault-tolerant).   L You have confused Tandem's site-disaster-tolerant approach (process pairing)L with fault tolerance.  Tandem's fault-tolerance is hardware-implemented, butC (as I noted before) hardware fault-tolerance isn't practical acrosseE site-disaster-tolerant distances, so they use a more coarsely-grained J (process-pair) mechanism for that purpose (which is high availability, not fault-tolerance).a   >iK > Note that shared memory in a Wildfire between VMS instances theoreticallytK > allows software to become fault tolerant and a backup process to continuee ai  > transaction from another node.  L No more so than any other (e.g., message-passing) mechanism would.  And onceG again, since fault-tolerance implemented at the software level not onlyaI exposes the software to *major* additional complexity but also slows downiL the system significantly, it's not an attractive option (and becomes less soG over time as hardware prices fall and software development costs rise).   L Note also that it is *never* possible to address fault-tolerance in softwareH alone without fully emulating an approach such as pair-and-spare or TMR,F since those are the *only* ways to detect those hardware failures thatK produce incorrect results rather than immediately crash the system instancenJ they're running on - which, again, is why the hardware approach is usually the one taken.  H I suggest that you re-read the material already presented in this threadG (and/or external sources) until you understand it well enough not to be  inclined to respond again.   - bill   ------------------------------    Date: 27 Jan 2002 07:07:54 -0800 From: wingwong@witty.com (wing)a Subject: copy subdirectory< Message-ID: <873e96d6.0201270707.139a30c@posting.google.com>   I am new to openVMS.  ' Please let me raise a obvious question.   A How to copy a directory with sub-directories with a command?  Forc; example, if the directory structure of DIR_A is as follows.d   DIR_Ad  -DIR_AA    -DIR_AABi  -DIR_BB    -DIR_BBBl  D And I would like to copy the sub-directory (DIR_AA, _AAB, _BB, _BBB)4 to a directory, says DIR_B, how is the command like?   Thanks in advance.   SW   ------------------------------  + Date: Sun, 27 Jan 2002 16:22:10 +0100 (MET)w9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>g Subject: Re: copy subdirectory; Message-ID: <01KDKLQQU5HQ8ZIVW3@sysdev.deutsche-boerse.com>a  @ > How to copy a directory with sub-directories with a command?   > F > And I would like to copy the sub-directory (DIR_AA, _AAB, _BB, _BBB)6 > to a directory, says DIR_B, how is the command like?  " $  COPY [DIR_A...]*.*;* [DIR_B...]   ------------------------------    Date: 27 Jan 2002 09:46:17 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)n Subject: Re: copy subdirectory3 Message-ID: <aiqzAMjyUsq+@eisner.encompasserve.org>o  ^ In article <873e96d6.0201270707.139a30c@posting.google.com>, wingwong@witty.com (wing) writes: > I am new to openVMS. > ) > Please let me raise a obvious question.t > C > How to copy a directory with sub-directories with a command?  Fore= > example, if the directory structure of DIR_A is as follows.t >  > DIR_As
 >  -DIR_AA
 >    -DIR_AABs
 >  -DIR_BB
 >    -DIR_BBB. > F > And I would like to copy the sub-directory (DIR_AA, _AAB, _BB, _BBB)6 > to a directory, says DIR_B, how is the command like?    $ 	$ DEFINE/USER SOURCE _ddcu:[DIR_A.]$ 	$ DEFINE/USER TARGET _ddcu:[DIR_B.]% 	$ BACKUP SOURCE:[*...] TARGET:[*...]m  I is approximately what I use, with your device name substituted for _ddcu. C I have not tested the above, and there may be something easier withn newer versions of VMS.   ------------------------------  + Date: Sun, 27 Jan 2002 16:52:29 +0100 (MET)s9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>  Subject: Re: copy subdirectory; Message-ID: <01KDKMT276SA8ZIVW3@sysdev.deutsche-boerse.com>o  & > 	$ DEFINE/USER SOURCE _ddcu:[DIR_A.]& > 	$ DEFINE/USER TARGET _ddcu:[DIR_B.]' > 	$ BACKUP SOURCE:[*...] TARGET:[*...]u > K > is approximately what I use, with your device name substituted for _ddcu.uE > I have not tested the above, and there may be something easier withy > newer versions of VMS.  F Note that COPY has been able to do [...] for a LONG time now; no need  for BACKUP.o  G Note that COPY and BACKUP have different policies regarding timestamps.   H Neither with BACKUP nor with COPY is it NECESSARY to use logicals (just + trying to avoid confusing the newbie here).o   ------------------------------    Date: 27 Jan 2002 10:34:24 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)1 Subject: Re: copy subdirectory3 Message-ID: <k65ctkWg8519@eisner.encompasserve.org>t  w In article <01KDKMT276SA8ZIVW3@sysdev.deutsche-boerse.com>, Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> writes:u' >> 	$ DEFINE/USER SOURCE _ddcu:[DIR_A.]a' >> 	$ DEFINE/USER TARGET _ddcu:[DIR_B.]s( >> 	$ BACKUP SOURCE:[*...] TARGET:[*...] >> sL >> is approximately what I use, with your device name substituted for _ddcu.F >> I have not tested the above, and there may be something easier with >> newer versions of VMS.$ > H > Note that COPY has been able to do [...] for a LONG time now; no need 
 > for BACKUP.e  D I had private email from somebody yesterday who had problems because! COPY will not create directories.m  I > Note that COPY and BACKUP have different policies regarding timestamps.    Please elaborate.   J > Neither with BACKUP nor with COPY is it NECESSARY to use logicals (just - > trying to avoid confusing the newbie here).f  J I was not confident of a proper command string for BACKUP without testing. Perhaps you can provide one.   ------------------------------  # Date: Sun, 27 Jan 2002 14:03:54 GMT ? From: Jim.Johnson@software-exploration.nospam.com (Jim Johnson) < Subject: Re: DECdtm DTI$S_PART_NAME should be (no *IS*) 255!/ Message-ID: <3c54050d.9511366@news.demon.co.uk>p   Richard,  F I've been trying to figure out what to suggest you do.  I do know thatE I'm not the correct person to speak to, for any number of reasons.  IsC suspect that its Rick McLaughlin, but I'd like to offer some advice D first.  Please note that I've not talked with anyone at Compaq aboutF this, and that this is based on thinking about your request from first principles.A  E First, I would expect that he's going to want information on specificcE revenue opportunities that absolutely require this change.  Since yout? can build what I'm guessing is the gateway you're interested in A without these changes (albeit with some additional complexity and ? perhaps some performance), I suspect that is going to be tough.a; However, without specific instances I doubt you'll succeed.r  B Second, to put this in some perspective, the changes would requireF modifications to the existing base services.  While those changes are,A in and of themselves, not large, their placement and implicationsoE would require a) that they be included in a base system update -- not.: something that comes along every week, and b) an extensiveD revalidation and load testing.  This is one of those few week coding tasks with a long testing tail.t  D Now, I'm sure you're going to ask why the revalidation would need toD be so extreme.  It's simple: the load on the logging subsystem wouldA be capable of being increased dramatically (by using lots of much C larger RMs).  The first obvious questions are: Does that induce anymF new problems in the logger?, Do transactions with large RMs induce anyF performance penalties (e.g. during fuzzy checkpointing) on 'old-style'B transactions?, Do the tools and other parts of the existing DECdtmF code handle all the edge conditions for long RM names?  While I'm sureE that the changes could be made such that it wouldn't be a problem, itH< would be wrong of OpenVMS not to insist upon verifying that.  D Remember, the XA work could be released quickly because it was fullyB layered on the existing service, not modifications inside of them.  E So, I'm not saying it can't be done.  I am saying that I suspect, and C I could be wrong, that you will find that you need to make a solid,aF verifiable financial case for OpenVMS putting its resources into this.    D As for the takeup on the XA kit, I honestly have no idea what it has been.t   Jim.    9 On Wed, 23 Jan 2002 08:50:25 +0000 (UTC), "Richard Maher"g <maher_rj@hotmail.c0m> wrote:f   >Hi Jim, >h >Thanks for replying.  >o1 >>and unless the participant size limit proves tobE >>be a killer for someone doing something like that for a popular 2PCi@ >>protocol/api, I don't realistically see it getting a very high
 >>priority >cM >Simply stated, without this functionality I'm dead in the water. I too don't K >want to go into to much detail in this forum as to what I plan to stick innL >the participant name and what it will be used for. (I'm sure you've already% >worked it out but others won't have)  >tE >>I suspect most of those will be building interoperability gateways.I >>That can now be done >nE >Yes that can be done now but not with the performance, transparency,tJ >manageability and reduced footprint that is achievable with a new $getdtiH >item code/flag. (In my version 1.0 developers will only be able to join! >transactions, not initiate them)O >oD >Should I mail you and Rick McLaughlin directly with a more detailedC >explanation/justification for this change or are there more formaln >procedures required?r >rI >I'm not going to badger with how easy and quick it would be to make thisiL >change. This simple fact is I don't know and you do but given the nature ofK >the change required and the fact that that the $getdti/$setdti bits appear J >to live in a seperate shareable then, as you've already pointed out, it's >not rocket science. >'C >>Aside from the time it would take to get this change out into ther >>customer hands,o > H >I know you've got a whole lot of regression testing to do but would you$ >really have to get Oracle involved? >w$ >How about a Christmas release date? >dH >>First, thanks for your appreciation for the EFT kit and documentation.F >>I'm glad that someone has taken the trouble to download and read it. >nM >Has take up been disappointing? I know of one very large French company thatlJ >was gagging for a Tuxedo XA interface to Rdb not so long ago. Could be an0 >ideal reference site? Paris in the Spring time? >mE >And please don't forget the branch management services! These shouldoJ >*always* have been released with the other public system services. I haveJ >some Cobol (yes that is what your customers use!) examples of transactionI >propagation, one a DECnet task-to-task and the other a sub-process. BothfM >updating Rdb in parrallel and both examples very small. They're yours if youe >want them.o > L >There was this guy in the Rdb list server that had a transaction spanning aL >combination of 40 Rdb tables and RMS files. I suggested some load balancingH >followed by mutli-processing and gave hime the examples but never heard3 >back. If you would like his name then let me know.e >o >Cheers Richard. >eJ >PS. $*_rm seem to let me put in 256 characters (as per the old functionalM >spec) but LMCP seems to have trouble dumping over 128 and its hex dump seems J >to want somewhere around a hundred? The best you can will be good enough. >i >Jim wrote:a >vG >First, thanks for your appreciation for the EFT kit and documentation.bE >I'm glad that someone has taken the trouble to download and read it.i > E >Now, for the rest of this, I do have to consider what I say in light-D >of a) things I've signed, b) things Software Exploration may or mayB >not wish to do with Compaq in the future, and c) the fact that it >ain't my product... >v >3B >The restriction on the participant name reflects the maximum sizeG >fully supported by DECdtm.  As you noted, there is some latent supportnA >for longer names in there, but you should consider it just that. > >Specifically, as you noticed, that support does not extend to
 >recovery. >tA >You noted that you can see the longer names using LMCP.  That is E >because, in these cases, LMCP accesses the log directly.  It does so G >for reasons both historical (it was built simultaneously with $GETDTI)uC >and practical (it needs to work under conditions where the log maye% >have seen some level of corruption).  >lC >Finally, there's the question of whether or not the latent support E >could be completed.  As software is, by it's very nature, soft, then @ >yes, it could.  Personally, I might even find it a smallish andA >localized change.  However, it would require changing a core VMSTG >component, and would therefore need to go through a much more involvedl@ >process than the XA work did (as that was layered upon existing
 >components).  > B >Aside from the time it would take to get this change out into theC >customer hands, the real issue is one of priority.  The reality isoF >that relatively few people will use the DECdtm services directly, andD >I suspect most of those will be building interoperability gateways.F >That can now be done, and unless the participant size limit proves toD >be a killer for someone doing something like that for a popular 2PC? >protocol/api, I don't realistically see it getting a very high A >priority.  However, that is a strictly personal opinion, and ther1 >decision is not, and should not be, in my hands.a >d >Jim.g >t >M > 9 >On Tue, 8 Jan 2002 23:20:45 +0000 (UTC), "Richard Maher"a# ><maher_rj@notsohotmail.com> wrote:n >  >>Hi,n >>C >>In the OpenVMS DECdtm Services Reference Manual V2.0 6-Jun-01 TheeL >>Revision History for Jun 6th it continues "Reduce the size of part_name toF >>match dti$t_part_name". This was clearly a mistake! What should have >changed? >>is dti$s_part_name and dti$t_part_log_id and so on and so on.l >>L >>Ok I know this is not possible because it would break existing code so canJ >>we _please_ have either a new item code like dti$_txn_info_full or a newA >>flag dti$m_search_verbose or dti$m_full_part_name or dti$m_v21?, >>L >>Only the output list need change. It's fair to say that different productsI >>should have part_names unique within the first 32 characters so current   >>input wildcarding can prevail. >>L >>It's only $getdti that needs to change. I am reliably informed that it's aL >>particularly nasty piece of code, and the current DECdtm developers appearL >>loath to touch any of the underlying services but you know it makes sense? >>G >>How can $join_rm currently let me enter > 32 byte part_names and LMCP J >>happily display them if there was a serious restriction? Surely this hasH >>always been a buglet with $getdti that can't be fixed directly (upwardK >>compatibility - I always said that you released the API documentation too . >>soon :-) but a new item code has to be easy! >>J >>I don't wish to appear ungrateful! The System Service API docs have beenG >>made publicly available and XA compliance has been introduced. Pleasea >acceptfI >>my sincere and heartfelt thanks! But I'm begging ya for this (probably)  >lastf >>(probably) little thing. >>E >>If Jim and crew have moved on to something else I know someone witheE >>experience in debuging $getdti who might be interested in doing it.t >>K >>If the answer is no then is there another API for getting at the txn log?l/ >>The truth and the real part_name is in there!s >> >>Regards Richard Maher. >> >> >> >> >> >> >> >> >* >Jim Johnson >Software Exploration, Ltd.t* >(remove '.nospam' from the reply address) >f >s >o >e >o >e >o   Jim Johnson  Software Exploration, Ltd.) (remove '.nospam' from the reply address)a   ------------------------------  % Date: Sat, 26 Jan 2002 12:13:31 -2000  From: mardell@desertmail.com& Subject: http://birthdaybill.81832.com$ Message-ID: <023cc4214201a12S11@s11>   http://birthdaybill.81832.com     % Mortgage rates have NEVER been lower.l  B Is your credit good?  Get a loan beyond your wildest expectations!  L Your credit stinks?  Lenders will still give you an absolutely amazing loan.  > Just click here and get started  http://birthdaybill.81832.com    ! Removal: http://www.e-mps.org/en/    ------------------------------  % Date: Sun, 27 Jan 2002 09:37:38 +0100e From: zessin@decus.deh* Subject: Re: I/O Performance on Alpha 4100* Message-ID: <00A08A9E.B8EBFBD1.1@decus.de>   Dirk Munk wrote: >BrianNFO wrote:I >>I haven't posted here in a while, but I'm looking to speed up I/O on an J >>Alpha 4100 5/400 running VMS 7.2-1.  Most of the devices are things likeL >>RZ40-VA, and I believe the VA indicates narrow scsi?  I don't know exactly" >>what PCI controllers are in use. >eE >Yes, VA stands for narrow, VW is wide. Grey/beige storage blocks aret* >fast SCSI, blue storage blocks UltraWide.  A Grey/beige? Hm. I have been told a story about someone who failed A a test because the color was called 'tan' in a question. However,n7 the unfortunate person thought they meant 'ten' (10)...o  B Oh, and I have seen DS-RZ1CF-VA and DS-RZ1DF-VA disks in a top-gun
 blue carrier.t   -- n
 Uwe Zessin   ------------------------------  % Date: Sun, 27 Jan 2002 10:41:29 -0500n0 From: "Island Computers" <dbturner@islandco.com>* Subject: Re: I/O Performance on Alpha 4100/ Message-ID: <u587siqkj8uu51@news.supernews.com>f   I beleive the colors are:c   Frost White - Prioris ) Nantucket Gray  - Alpha Servers and Diskst5 Top Gun Blue - All new stuff before the Charcoal Grayn   ------------------------------  % Date: Sat, 26 Jan 2002 08:39:41 -0500o+ From: "Main, Kerry" <Kerry.Main@Compaq.com>  Subject: RE: Kerberos in VMST Message-ID: <BE56C50EA024184DAF48F0B9A47F5CF401AB1BEC@kaoexc01.americas.cpqcorp.net>   Neil,a   Fyi -sF http://www.openvms.compaq.com:8000/73final/6647/kerberos_relnotes.html   Regardse  
 Kerry Main Senior Consultantl Compaq Canada Corp.n Professional Services  Voice: 613-592-4660t Fax  :  819-772-7036 Email: Kerry.Main@Compaq.com     -----Original Message-----5 From: Neil Hemingway [mailto:hemingway_n@hotmail.com]: Sent: January 25, 2002 6:48 PM To: Info-VAX@Mvb.Saic.Comc Subject: Kerberos in VMS    C In message <8m9bn4$39g$3@mailint03.im.hou.compaq.com>, Hoff Hoffmanc wrotei  H >   The next protocol being added for external authentication support isE >   Kerberos (both the client and server), and this is expected to be  part of  >   the OpenVMS V7.3 release.e  E 7.3 is now out, and I haven't found any mention of Kerberos in it.=20sF Did it slip from this release, or am I looking in completely the wrong place?   ------------------------------  # Date: Sun, 27 Jan 2002 13:40:19 GMTl1 From: Ed Wensell III <ewensell3@yahoo.commercial>u Subject: Re: Makefilec0 Message-ID: <3C5402EF.3DEEE722@yahoo.commercial>   cljlk wrote: > H > I am new in VAX/VMS system and don't know how to create a makefile for
 > C programs.t> > Would anyone please send me a makefile, thus, I can edit it?> > Please give me an instructing how to use it on command line. > Thanks in advance.  F FWIW, most C sources that include a VMS port provide a (hand created?)E build.com to build the VMS version. Makefiles are typically UNIXisms. H Most source packages for VMS assume make is not available on the system.  C The latest trend in source distribution on UNIX machines is to alsoiA include a 'configure' shell script which tweakes the Makefile anduD config.h for you based on what resources are available on the systemC (which C compiler, graphics libraries, processor specific commands,lG etc). This makes build slightly more reliable. Something similar in VMSB0 would be nice, but would probably be impossible.   -- _ Ed Wensell IIIE NetBSD/Alpha at home - Solaris/SPARC at work - OpenVMS in a past lifeFA E-mail address is valid if you know the appropriate bits to drop.]   wibble?    ------------------------------  + Date: Sun, 27 Jan 2002 17:56:48 +0000 (UTC)e From: david20@alpha2.mdx.ac.uk Subject: Re: Makefilea+ Message-ID: <a31f10$9p5$1@aquila.mdx.ac.uk>   U In article <3C532EA3.58A9F4BB@pacbell.net>, Don Sykes <anonymous@pacbell.net> writes:0 >M >s
 >cljlk wrote:  >> gI >> I am new in VAX/VMS system and don't know how to create a makefile fors >> C programs.? >> Would anyone please send me a makefile, thus, I can edit it?C? >> Please give me an instructing how to use it on command line.C >> Thanks in advance.TG >I believe there is a GNUmake available for VMS you can use. Other thantC >that, VMS has a facility called MMS, which is similar to make, buteH >you'll need a license. Or you can roll your own - cc + link statements. >t    8 There is also a public domain version of MMS called MMK.   see   % ftp://ftp.process.com/madgoat/mmk.zipa     From the AAAREADME.DOC file :-   " O MadGoat Make Utility (MMK)                                                     hO                                                                                -O MMK is a "make" utility for VMS systems.  It is used for building software     oO systems based on a "description file" (or "makefile") you create that lists    6O the sources and objects of a system and the dependencies between them.         rO                                                                                GO MMK is similar in functionality to Digital's DEC/Module Management System (MMS)mO and understands a syntax in its description files which is mostly compatible   tO with MMS.                                                                       O                                                                                oO MMK runs on VAX/VMS, OpenVMS VAX, and OpenVMS AXP systems.  It requires        eO VAX/VMS V5.2 or later, or OpenVMS AXP V1.0 or later.                           tO                                                                                oO MMK is written entirely in C.  Complete source code is provided.               hO                                                                                h "   
 David Webb VMS and unix team leader   CCSS Middlesex University   ------------------------------  # Date: Sun, 27 Jan 2002 09:03:35 GMTp" From: "Hans Vlems" <hvlems@iae.nl>& Subject: Re: mount problems on cluster0 Message-ID: <HnP48.1431$A3.7139@typhoon.bart.nl>  B First of all, mounting disks in sylogicals.com is not a good idea.F If something happens to your diskfarm, and you do need to do a minimalJ boot then you're in trouble since sylogicals is always executed. There are- ways around that, but why use sylogicals.com?n  6 What happens if you put "sho dev d" in sylogicals.com?   Hans  < Robert DiRosario <rdirosario@starpower.net> wrote in message' news:3C535DF6.5789CF41@starpower.net...'3 > I'm having problems mounting disks on my cluster.o >n. > If I reboot a node and then do "show dev /m"5 > some of my DSSI disks show up with "(remote mount)"r2 > listed for the volume label.  If I try to access0 > the disk I get errors.  I can access the disks > just fine from other nodes.i >t- > If I issue the mount commands on a terminale > everything works fine. > 3 > If I power down everything and boot all the nodesw& > from scratch, everything works fine. > 4 > Why are the DSSI disks not being mounted correctly > from sylogicals.com? >p > In sylogicals.com I have:o > - > $mount $2$dia1  disk_d1  /cluster /noassiste- > $mount $2$dia2  disk_d2  /cluster /noassiste- > $mount $2$dia10 disk_d10 /cluster /noassistl- > $mount $2$dia11 disk_d11 /cluster /noassisti- > $mount $2$dia12 disk_d12 /cluster /noassistm- > $mount $2$dia13 disk_d13 /cluster /noassistw >d2 > $mount m48$dka200: disk4      /cluster /noassist2 > $mount m48$dkb200: disk1      /cluster /noassist2 > $mount m48$dkb300: disk2      /cluster /noassist2 > $mount m48$dkb500: disk3      /cluster /noassist > $ > The cluster system disk is $2$dia0 >n, > I have nodes A (4000/106A), B (4000/106A),/ > C (4000/105A) and M48 (VAX Station 3100/M48).' >-. > Nodes A, B, and C, disks dia0, dia1 and dia2" > and a HSD10 are on one DSSI bus. > 4 > Nodes A and B, disks dia10, dia11, dia12 and dia13+ > and a 2nd HSD10 are on a second DSSI bus.t > 2 > Node C and disks D20, D21 and D22 are on a third2 > DSSI bus, but I haven't added mount commands for > those drives yet.l >e2 > Node m48 is on ethernet and works fine, but it's6 > turned off for now.  It has it's page and swap files5 > on a local SCSI disk, and when nodes A, B or C boote6 > they run sypagswpfiles.com and try to use m48's page4 > and swap file.  (A quick search of google turns up6 > satellite_page.com as a solution for old versions of6 > VMS, but I'm using 7.1.  Is the node name defined by# > the time sypagswpfiles.com runs?)n >0 > Thanks > Robert   ------------------------------  % Date: Sun, 27 Jan 2002 08:13:48 -0500k2 From: norm lastovica <norman.lastovica@oracle.com>& Subject: Re: mount problems on cluster* Message-ID: <3C53FD0C.8221C683@oracle.com>  9 I'll bet that the configure process isn't running at thatg> point in time.  Why not put the disk mounts into SYSTARTUP_VMS9 instead of SYLOGICALS?  configure will be running at thatn( point and all the disks will be visible.   Robert DiRosario wrote:p > 3 > I'm having problems mounting disks on my cluster.o > . > If I reboot a node and then do "show dev /m"5 > some of my DSSI disks show up with "(remote mount)"p2 > listed for the volume label.  If I try to access0 > the disk I get errors.  I can access the disks > just fine from other nodes.i > - > If I issue the mount commands on a terminaln > everything works fine. > 3 > If I power down everything and boot all the nodesn& > from scratch, everything works fine. > 4 > Why are the DSSI disks not being mounted correctly > from sylogicals.com? >  > In sylogicals.com I have:o > - > $mount $2$dia1  disk_d1  /cluster /noassistr- > $mount $2$dia2  disk_d2  /cluster /noassisti- > $mount $2$dia10 disk_d10 /cluster /noassistp- > $mount $2$dia11 disk_d11 /cluster /noassiste- > $mount $2$dia12 disk_d12 /cluster /noassisth- > $mount $2$dia13 disk_d13 /cluster /noassisti > 2 > $mount m48$dka200: disk4      /cluster /noassist2 > $mount m48$dkb200: disk1      /cluster /noassist2 > $mount m48$dkb300: disk2      /cluster /noassist2 > $mount m48$dkb500: disk3      /cluster /noassist > $ > The cluster system disk is $2$dia0 > , > I have nodes A (4000/106A), B (4000/106A),/ > C (4000/105A) and M48 (VAX Station 3100/M48).s > . > Nodes A, B, and C, disks dia0, dia1 and dia2" > and a HSD10 are on one DSSI bus. > 4 > Nodes A and B, disks dia10, dia11, dia12 and dia13+ > and a 2nd HSD10 are on a second DSSI bus.v > 2 > Node C and disks D20, D21 and D22 are on a third2 > DSSI bus, but I haven't added mount commands for > those drives yet.? > 2 > Node m48 is on ethernet and works fine, but it's6 > turned off for now.  It has it's page and swap files5 > on a local SCSI disk, and when nodes A, B or C bootc6 > they run sypagswpfiles.com and try to use m48's page4 > and swap file.  (A quick search of google turns up6 > satellite_page.com as a solution for old versions of6 > VMS, but I'm using 7.1.  Is the node name defined by# > the time sypagswpfiles.com runs?)  >  > Thanks > Robert   -- b> norman lastovica / oracle rdb engineering / usa / 610.696.4685   ------------------------------  # Date: Sun, 27 Jan 2002 14:12:45 GMTy- From: "John E. Malmberg" <wb8tyw@qsl.network>t& Subject: Re: mount problems on cluster* Message-ID: <3C540F0D.9080300@qsl.network>   Robert DiRosario wrote:o  3 > I'm having problems mounting disks on my cluster.l > . > If I reboot a node and then do "show dev /m"5 > some of my DSSI disks show up with "(remote mount)" 2 > listed for the volume label.  If I try to access0 > the disk I get errors.  I can access the disks > just fine from other nodes.m > . > If I issue the mount commands on a terminal  > everything works fine. > 4 > If I power down everything and boot all the nodes & > from scratch, everything works fine. > 4 > Why are the DSSI disks not being mounted correctly > from sylogicals.com?      C Usually disk mounts are done at the SYSTARTUP_VMS.COM phase of the rH startup.  Use SYLOGICALS.COM for system wide logical names needed after 	 the boot.d  F When the system is being booted from a power up, it can take some timeF for all of the MSCP devices to be discovered.  Until then they can not be mounted.r  D So basically at boot time, you will need a to have a loop that uses C f$getdvi(devname,"exists") with a delay of about 5 seconds between sI tries.  It should also only try for about 12 times, so that if something i2 is wrong with the disk, the startup will not hang.  I Using the /CLUSTER on a mount command in the SYSTEM BOOT sequence is not cI recommended.  You do not know if the other nodes in the cluster are able n to service the request.p  I If you are serving disks back from a diskless satellite, then use of the rE /CLUSTER will usually work, because typically the other hosts in the eH network are already up.  You may still have to do some manual mounts if F you have multiple nodes booting at the same time, but usually it will I come out right.  About the only way to have it always work would involve eF setting up a detached process or a batch job that makes sure that all  the disks got mounted.   See SYS$EXAMPLES:MSCPMOUNT.COM.e  H Also see http://encompasserve.org/~malmberg/dcl/ for MOUNT_DISKS.TXT and MOUNT_DISKS_DRAGON.TXT.n  D Mount_disks.txt looks up all the systems currently in a cluster and K executes the appropriate STARTUP_DIR:MOUNT_DISKS_*.COM to get them mounted.f  < MOUNT_DISKS_DRAGON.TXT mounts the disks for a specific node.    2 > Node m48 is on ethernet and works fine, but it's7 > turned off for now.  It has it's page and swap files u6 > on a local SCSI disk, and when nodes A, B or C boot 7 > they run sypagswpfiles.com and try to use m48's page a4 > and swap file.  (A quick search of google turns up6 > satellite_page.com as a solution for old versions of6 > VMS, but I'm using 7.1.  Is the node name defined by# > the time sypagswpfiles.com runs?)c    I I would recommend using f$edit(f$getsyi("SCSNODE"),"UPCASE,TRIM") to get sE the system name.  The SCSNODE name is available to all user editable I command files in the boot.  I Satellite_page.com is generated by the CLUSTER_CONFIG.COM procedure when tG you configure a satellite system.  It is used to mount the remote disk d' and install a page and swap file on it.>  E In this case you would not use the SYPAGSWPFILES.COM unless you were s2 adding additional page or swap files for the node.  F Note that SYPAGSWPFILES.COM is executed even for minimum startups, so F when I use one, I put some logic in it to prevent it from running, as E when I am booting minimum, I may need to do work on the disk(s) that r0 have the additional page and swap files on them.  ' See SYPAGSWPFILES.TXT at the above URL.   H Because the disk for the secondary page and swap files may not be ready,G The SYPAGESWPFILES procedure must also poll and wait for the disk to beb configured and placed on line.  K To prevent code duplication, I have the SYPAGSWAPFILES_<node>.COM file calleM the appropriate STARTUP_DIR:MOUNT_DISKS_<node>.COM with the parameter "PAGE".>      E To help understand the command files that I posted on Encompasserve,  G please note that in general, I use a naming convention for files on my   systems as follows:e  > No suffix: The file can be placed on any system that I manage.  9 A suffix of "_<node>": The file is specific to that node.m  D A suffix of "_<cluster_alias>_COMMON.COM:  This file is common to a , group of nodes that share the cluster_alias.  E This allows me to keep all of the files in the SYS$COMMON: directory t" tree, and in a common CMS library.   -Johni wb8tyw@qsl.network   Personal Opinion Only>   ------------------------------  % Date: Sat, 26 Jan 2002 07:38:13 -0500t+ From: "Main, Kerry" <Kerry.Main@Compaq.com> 1 Subject: RE: new partner also sent to openvms.orgeT Message-ID: <BE56C50EA024184DAF48F0B9A47F5CF401AB1BEB@kaoexc01.americas.cpqcorp.net>  G And to just add to Sues posting .. As mentioned before, this is a greataA example where an application written in Java becomes very easy to2 certify on OpenVMS.t  , Check out http://www.imsure.com/Home.html=20   Regards,    
 Kerry Main Senior Consultant? Compaq Canada Corp.o Professional Services! Voice: 613-592-4660s Fax  :  819-772-7036 Email: Kerry.Main@Compaq.com     -----Original Message----- From: Skonetski, Susan=20i Sent: January 25, 2002 4:09 PM To: Info-VAX@Mvb.Saic.Com-- Subject: new partner also sent to openvms.org     D IMSure Network, Inc. is a technology company that provides eCommerceG Internet technology products that make businesses more profitable.   WerE provide eCommerce infrastructure services, enterprise level software,k andoA end-2-end eCommerce service provider (ESP) support. Our products,f B2BflexcC and B2Cflex, enable eCommerce applications to be deployed, managed,sG supported and upgraded from centrally located data centers, rather than  on individual company servers.   G IMSure Network's  B2B/B2C e-commerce package is designed for hosting in G multiple cross-platform environments.  The package was written entirelye inF Java for that very reason.  In mid-summer 2001, it was decided to port thisG package to the OpenVMS environment.  Ensuring that the software package  wasSH pure Java resulted in an application that is easy to run on any properlyE implemented Java virtual machine.  The key result is the applicationsmB ability to run on a wide variety of platforms with ease, including almostG all operating systems and platforms currently available.  When platformvD independence is needed, Java is clearly an outstanding technology to work with.i       Regards, Mark Muehleisenf markm@imsurenetwork.comn iMSure Network, Inc. 480.456.3811 x 205   ------------------------------  % Date: Sun, 27 Jan 2002 12:10:51 -0500c2 From: rdeininger@mindspring.com (Robert Deininger)1 Subject: Re: Northernlight to close public accessiJ Message-ID: <rdeininger-2701021210520001@1cust91.tnt3.nashua.nh.da.uu.net>  O In article <00A08074.CD532E0E@SendSpamHere.ORG>, system@SendSpamHere.ORG wrote:p  D >In article <0hm54u4qu0elogh0438move3g6ebfal2l4@4ax.com>, Alan Greig <a.greig@virgin.net> writes: >>G >>The VMS based search engine Northern Light (www.northernlight.com) ismH >>to close its public search facility. A real pity because it was a nice7 >>publicity tool as well as an excellent search engine.46 >>http://www.northernlight.com/docs/letterfromceo.html  L >So they are going from an advertisement-supported model to what, a customer3 >supported model?  If so, how does one subscribe?  s  F Last year, at an OpenVMS forum, someone from NorhternLight presented aH "Why we use OpenVMS" talk.  He said that web ads were not a major sourceF of their income.  He mentioned two other areas that brought in most of
 their income:-- 1) Fee-based searches on the public web site.MI 2) Contract-based indexing and search services within a company's privatelH network.  NL could either host the searches on their servers, or provideC software to run on the other company's server.  This sounded like afH near-custom service, with minor (or major) tweeks to meet each company'sG needs.  I did not get the impression they had a shrink-wrapped software  product.  J He was happy that advertising revenue was not too important to them, since> the era of dot.bomb had made that a very risky business model.   ------------------------------  % Date: Sun, 27 Jan 2002 09:37:23 -0800n# From: "Tom Linden" <tom@kednos.com>o* Subject: passing /debug to linker from MMS9 Message-ID: <CIEJLCMNHNNDLLOOGNJIAEGIEAAA.tom@kednos.com>k   In my description file I have:  L LINKFLAGS       = $(MAPFLAG)/section_binding=(code,data)/nodebug/notraceback   which the used ini  :  $(link)/SHARE=$(SHRLIB)$(LINKFLAGS) $(OPTIONFILE)/options  K I would like to put in /debug/cross/map.  Is there a way to pass these from L the mms command?  How should I modify the description file to override thoseI qualifiers?  The MMS command is executed from within a script, This is on  Alpha  6.2, 7.1-2 and 7.3   TIA- Tom-   ------------------------------  # Date: Sun, 27 Jan 2002 18:48:31 GMT@- From: "John E. Malmberg" <wb8tyw@qsl.network>g. Subject: Re: passing /debug to linker from MMS* Message-ID: <3C544FA3.3040902@qsl.network>   Tom Linden wrote:l    > In my description file I have: > N > LINKFLAGS       = $(MAPFLAG)/section_binding=(code,data)/nodebug/notraceback >  > which the used ind > < >  $(link)/SHARE=$(SHRLIB)$(LINKFLAGS) $(OPTIONFILE)/options > M > I would like to put in /debug/cross/map.  Is there a way to pass these fromIN > the mms command?  How should I modify the description file to override those
 > qualifiers?t       LINKFLAGS = J $(MAPFLAG)/section_binding=(code,data)/nodebug/notraceback$(LINKDEBUGFLAG)  ? >  The MMS command is executed from within a script, This is ono > Alphac > 6.2, 7.1-2 and 7.3    I Then you can either have a DCL symbol LDEBUG = "/DEBUG/TRACEBACK" or add  # a line like that to the MMS script.   E I am guessing that $(MAPFLAG) is used to determine if a MAP is to be  I produced.  If MAPFLAG is not defined inside of your MMS script, then you t- can have a DCL SYMBOL MAPFLAG = "/CROSS/MAP".t     C With MMS, if a macro name is not defined in the MMS script, then iteE checks to see if a DCL symbol exists with that name, and will use it.i    C This means you must be careful with the names of macros in the MMS  3 script, as some interesting side effects can occur.    -Johnl wb8tyw@qsl.network Personal opinion onlye   ------------------------------  % Date: Sun, 27 Jan 2002 12:52:13 -0600 C From: "Craig A. Berry" <craig.berry@nospam.signaltreesolutions.com> . Subject: Re: passing /debug to linker from MMS= Message-ID: <3C544C5D.6030708@nospam.signaltreesolutions.com>M   Tom Linden wrote:e    > In my description file I have: > N > LINKFLAGS       = $(MAPFLAG)/section_binding=(code,data)/nodebug/notraceback >  > which the used in  > < >  $(link)/SHARE=$(SHRLIB)$(LINKFLAGS) $(OPTIONFILE)/options > M > I would like to put in /debug/cross/map.  Is there a way to pass these from N > the mms command?  How should I modify the description file to override thoseK > qualifiers?  The MMS command is executed from within a script, This is onr > Alpha 6.2, 7.1-2 and 7.3    G I suspect you'd want to add another macro and use the /MACRO qualifier s on the command line:   $ type descrip.mms target :;          @ write sys$output "The value of foo is: '$(foo)'" " $ mms/macro=foo="/debug/cross/map"  ' The value of foo is: '/debug/cross/map'o   ------------------------------  % Date: Sun, 27 Jan 2002 11:51:33 -0500i2 From: rdeininger@mindspring.com (Robert Deininger)! Subject: Re: Selling VMS licensescJ Message-ID: <rdeininger-2701021151330001@1cust91.tnt3.nashua.nh.da.uu.net>  < In article <e9cbc4f2.0201161124.c60d97c@posting.google.com>,- mcbill20@hotmail.com (Bill McLaughlin) wrote:i    F >I still have this computer but I am running Oracle and need more thanE >the 128mb the box came with. Unfortunately memory (MSP01-FC, -FD) is) >$400 for an additional 128mb. .  E For cheaper alpha memory upgrades, consider buying used memory and/or  3rd-party memory.r  I I have had very good experiences with memory from Clearpoint on a varietyr9 of alpha (and Vax) systems.  You can find contact info at J www.clearpoint.com . There are various other vendors who can supply memory for alpha systems.   ------------------------------   Date: 27 Jan 2002 18:34:09 GMT( From: seebs@plethora.net (Peter Seebach)Y Subject: Re: Sv: Sv: Sv: Younger recruits versus experienced veterans ( was Re:     The de9 Message-ID: <3c544821$0$79557$3c090ad1@news.plethora.net>i  * In article <3c47b82f.98580@news.ocis.net>,+ Gene Wirchenko <genew@mail.ocis.net> wrote:sF >     Two works fine for me.  I could handle three if the third buttonG >wasn't used much (like a wheel).  I'd probably want it to doubleclick.   J I am very happy with my current set of 3 buttons under twm; left to selectN and move, middle is maximize/unmaximize, right is close.  Then, with somethingM held down, they're "bring to front", "toggle front/back", and "push to back",, I believe.  Good enough for me.   * >seebs@plethora.net (Peter Seebach) wrote:M >>It seems to be - past that, I am starting to lose track of which buttons doI >>what.   3 >     Peter, you're not a touch-mouser?  Bad kitty!   F Oh, I am, but I don't use more than about 2-3 buttons, plus modifiers.   -s -- 8K    Copyright 2001, all wrongs reversed.  Peter Seebach / seebs@plethora.netpK    $ chmod a+x /bin/laden      Please do not feed or harbor the terrorists. J      C/Unix wizard, Pro-commerce radical, Spam fighter.  Boycott Spamazon!N Consulting, computers, web hosting, and shell access: http://www.plethora.net/   ------------------------------  % Date: Sun, 27 Jan 2002 09:21:46 +0100s. From: Dennis Grevenstein <dennis@pcde.inka.de> Subject: TCP/IP problems, Message-ID: <3C53B89A.40438543@pcde.inka.de>   Hi,r  = After switching off a VAX without a clean shutdown I get someu) strange error when I try to start TCP/IP.0- The system run OpenVMS 7.2-1 and TCP/IP 5.0A.f% I get this error when I start TCP/IP:o  7 gethostbyname: non-translatable vms error code: 0x182B2S %rms-e-rnf, record not found  9 I already found out that this means that the system couldb9 not find a host name. I reconfigured TCP/IP, but still ito does not work.* So I tried to  register the name manually:   TCPIP> show ho */ %TCPIP-E-HOSTERROR, cannot process host requestr( -TCPIP-W-NORECORD, information not found  -RMS-E-EOF, end of file detected  4 I get the same error when I try the register a name.   What's broken here?    TIAs Dennis   -- m? The long journey to world domination begins with a single step.d  ?                                                          Illiadw   ------------------------------  % Date: Sun, 27 Jan 2002 20:16:10 +1100m= From: "Mark\(unMASK\)Forsyth" <forsytMhm@optAushoSme.com.aKu>t Subject: Re: TCP/IP problems" Message-ID: <7hg03a.bpd.ln@really>  ; "Dennis Grevenstein" <dennis@pcde.inka.de> wrote in messages& news:3C53B89A.40438543@pcde.inka.de... > Hi,s >e? > After switching off a VAX without a clean shutdown I get some1+ > strange error when I try to start TCP/IP.m/ > The system run OpenVMS 7.2-1 and TCP/IP 5.0A.t' > I get this error when I start TCP/IP:t > 9 > gethostbyname: non-translatable vms error code: 0x182B2m > %rms-e-rnf, record not found >s; > I already found out that this means that the system couldn; > not find a host name. I reconfigured TCP/IP, but still itV > does not work., > So I tried to  register the name manually: >n > TCPIP> show ho *1 > %TCPIP-E-HOSTERROR, cannot process host requestd* > -TCPIP-W-NORECORD, information not found" > -RMS-E-EOF, end of file detected >o6 > I get the same error when I try the register a name. >  > What's broken here?n  H At a guess (I run UCX v ancient) I'd say the equivilent of ucx$host.dat.   Ooroou	 Mark F...u   >o > TIAn > Dennis >t > --A > The long journey to world domination begins with a single step.b >cA >                                                          Illiadp   ------------------------------  % Date: Sun, 27 Jan 2002 11:16:44 +0100i. From: Dennis Grevenstein <dennis@pcde.inka.de> Subject: Re: TCP/IP problems, Message-ID: <3C53D38C.92D55F46@pcde.inka.de>   "Mark(unMASK)Forsyth" wrote: > J > At a guess (I run UCX v ancient) I'd say the equivilent of ucx$host.dat.  > No. SYS$MANAGER:TCPIP$HOSTS.DAT and SYS$SYSTEM:TCPIP$HOSTS.DAT> contain the host name. BTW: Am I right that only the HOST file in SYS$SYSTEM really matters?.   Dennis   ------------------------------  % Date: Sun, 27 Jan 2002 22:21:29 +1100s= From: "Mark\(unMASK\)Forsyth" <forsytMhm@optAushoSme.com.aKu>d Subject: Re: TCP/IP problems" Message-ID: <6sn03a.kud.ln@really>  ; "Dennis Grevenstein" <dennis@pcde.inka.de> wrote in messageg& news:3C53D38C.92D55F46@pcde.inka.de... > "Mark(unMASK)Forsyth" wrote: > >IL > > At a guess (I run UCX v ancient) I'd say the equivilent of ucx$host.dat. >w@ > No. SYS$MANAGER:TCPIP$HOSTS.DAT and SYS$SYSTEM:TCPIP$HOSTS.DAT@ > contain the host name. BTW: Am I right that only the HOST file > in SYS$SYSTEM really matters?   L Hmm. That I can't answer - I don't have access to a box with better than UCX9 V4.2 - ECO 1 at the moment. FWIW on my system (hobby box)e' sys$system:ucx$host.dat is entered fromsK sys$sysdevice:[vms$common.sysexe]ucx$host.dat (both have the same file id).iK From that I'd guess (that word again) that the one in sys$system is the onea that really counts.   I Another Forsyth guess that didn't pan out, I wonder how I'll go with thisa	 one...:-)n  K Just had a thought. (yep, it did hurt). Why not try a $ set watch/class=allmK file followed by a $ tcp sho hos * the output may show you where it's goingU	 belly up.o   Ooroon	 Mark F...S   >O > Dennis   ------------------------------  # Date: Sun, 27 Jan 2002 13:05:01 GMTe- From: "John E. Malmberg" <wb8tyw@qsl.network>u Subject: Re: TCP/IP problems* Message-ID: <3C53FF2B.7020500@qsl.network>   Dennis Grevenstein wrote:s   > Hi,l > ? > After switching off a VAX without a clean shutdown I get somev+ > strange error when I try to start TCP/IP.i/ > The system run OpenVMS 7.2-1 and TCP/IP 5.0A.t' > I get this error when I start TCP/IP:a > 9 > gethostbyname: non-translatable vms error code: 0x182B2  > %rms-e-rnf, record not found > ; > I already found out that this means that the system couldM; > not find a host name. I reconfigured TCP/IP, but still it/ > does not work., > So I tried to  register the name manually: >  > TCPIP> show ho *1 > %TCPIP-E-HOSTERROR, cannot process host requestp* > -TCPIP-W-NORECORD, information not found" > -RMS-E-EOF, end of file detected > 6 > I get the same error when I try the register a name. >  > What's broken here?i    $ Your file TCPIP$HOST.DAT is corrupt.  E Compaq TCPIP uses the logical name TCPIP$HOST to find the file.  The  ' default location is SYS$COMMON:[SYSEXE]a  D The usual way that this file seems to get corrupted is from someone D using a text editor to modify it.  It would be unusual for a system C crash or other unclean shutdown to cause this file to be corrupted.w  G Delete the file, and then only use the TCPIP SET HOST to put your host "D names back in.  You can use a command file to add multiple hosts in.  E Most of the Compaq TCP/IP data files binary indexed files.  Any that mL were hand edited may need to deleted and recreated using the TCPIP commands.   -Johnt wb8tyw@qsl.network Personal Opinion Only    ------------------------------  % Date: Sun, 27 Jan 2002 15:50:01 +0100t. From: Dennis Grevenstein <dennis@pcde.inka.de> Subject: Re: TCP/IP problems, Message-ID: <3C541399.FB273F5D@pcde.inka.de>   "John E. Malmberg" wrote:a > E > The usual way that this file seems to get corrupted is from someonenE > using a text editor to modify it.  It would be unusual for a system E > crash or other unclean shutdown to cause this file to be corrupted.P  > You were right. I deleted these files and reran TCP/IP$CONFIG.A However I can't remember changing these files with a text editor.e; I have certainly done so after TCP/IP refused o work. Maybep: I have used Solaris too much. A text editor is normally my main administration tool.e   thanks Dennis   -- T Q: How can I open an iMac? A: I used a saw.   ------------------------------    Date: 27 Jan 2002 08:50:25 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)e< Subject: Re: VMS system on the sourceforge.net Compile Farm?3 Message-ID: <1fLTcEoVl6l7@eisner.encompasserve.org>s   In article <00A08A3E.3E5B468E@SSRL04.SLAC.STANFORD.EDU>, winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr") writes:iJ > In article <3C53870D.53EF8F2A@ACM.org>, "C.W.Holeman II" <cwhii@ACM.org> > w C >>Has anyone looked into having a VMS system on the sourceforge.netc >> Compile Farm? >  >>F >>	https://sourceforge.net/docman/display_doc.php?docid=762&group_id=1 > E > I looked up that page, and I see a couple of problems right away.  KK > The way the compile farm is set up you need SSH to log in and SCP to get eM > your code there and retrieve your binaries.  (User-id and password are alsoeN > fhe same as on other sourceforge projects, so although they don't explicitlyO > say it, they probably need YP/NIS or some other single-sign-in authenticationeO > xsystem, which may not run on VMS.)  Compaq has no SSH or SCP and no plans tonO > provide one; Multinet/TCPware have SSH and (I believe) SCP coming out in 4.4.2M > Although no money changes hands in the sourceforge project, it doesn't seeme > like legit hobbyist use.  C Without having read the information, it looks to me that it _would_ H qualify for a CSA system -- and there is precedent for other non-revenueE "vendors" of software.  It should just take someone willing to do thedC necessary politics, and I am sure the current VP of VMS Developmentn< would lend his support if it were properly explained to him.  M > These problems might not be insurmountable, but to really fix most of them,f< > either somebody would have to spend significant money or aJ > VMS-knowledgeable  person would to devote significant time to persuadingI > Compaq and Process to donate hardware and software, and then put in theaL > time and effort to keep the system software up to date (but with old RTLs,E > etc, available, so precompiled software can run on non-cutting edgeeJ > systems).  And even then, the people using it will probably be expectingM > autoconf to work.  So for the system to be useful, somebody's going to havetK > to be prepared to hold the hands of Linux weenies who never heard of VMS,yL > and who think (correctly) that they're doing VMS big people a big favor by > even trying to compile on it.p > N > That means that even a well-meaning Compaq couldn't just throw hardware and F > licenses at the problem, they'd have to devote a portion of a personM > (knowledgeable in both Unix and VMS) to it on an ongoing basis.  This seems8 > fairly unlikely.  G It doesn't seem that unlikely to me if there is someone willing to sellr VMS Development on the idea.   ------------------------------   Date: 27 Jan 2002 18:35:18 GMT( From: seebs@plethora.net (Peter Seebach)K Subject: Re: Younger recruits versus experienced veterans ( was Re: The demm9 Message-ID: <3c544866$0$79557$3c090ad1@news.plethora.net>0  + In article <a22p2e$3f2$1@panix3.panix.com>,n* Keith F. Lynch <kfl@KeithLynch.net> wrote:C >I still don't see the point in using a mouse at all, unless you'remA >using a paint program or playing a video game.  I never use one. A >I'm not sure how I *could* efficiently use one, since I use both> >hands for typing.  I I alternate modes.  If I'm browsing, the mouse is very useful for pickingsE an item from a list without having to wade through intervening items.oF Similarly, I like mice and menus for programs I only use occasionally.   -s -- eK    Copyright 2001, all wrongs reversed.  Peter Seebach / seebs@plethora.netrK    $ chmod a+x /bin/laden      Please do not feed or harbor the terrorists.hJ      C/Unix wizard, Pro-commerce radical, Spam fighter.  Boycott Spamazon!N Consulting, computers, web hosting, and shell access: http://www.plethora.net/   ------------------------------   End of INFO-VAX 2002.052 ************************ny's server.  This sounded like afH near-custom service, with minor (or major) tweeks to meet each company'sG needs.  I did not get the impression they had a shrink-wrapped software  product.  J He was happy that advertising revenue was not too important to them, since> the era of dot.bomb had made that a very risky businJ    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J     J    J    J    J    J    J    J    J    J    	J    
J    J    J    
J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J     J    !J    "J    #J    $J    %J    &J    'J    (J    )J    *J    +J    ,J    -J    .J    /J    0J    1J    2J    3J    4J    5J    6J    7J    8J    9J    :J    ;J    <J    =J    >J    ?J    @J    AJ    BJ    CJ    DJ    EJ    FJ    GJ    HJ    IJ    JJ    KJ    LJ    MJ    NJ    OJ    PJ    QJ    RJ    SJ    TJ    UJ    VJ    WJ    XJ    YJ    ZJ    [J    \J    ]J    ^J    _J    `J    aJ    bJ    cJ    dJ    eJ    fJ    gJ    hJ    iJ    jJ    kJ    lJ    mJ    nJ    oJ    pJ    qJ    rJ    sJ    tJ    uJ    vJ    wJ    xJ    yJ    zJ    {J    |J    }J    ~J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    J    