1 INFO-VAX	Fri, 24 Sep 2004	Volume 2004 : Issue 530       Contents:> Re: 1920x1200 resolution with VMS and the HP L2335 LCD monitorH Re: Backing up system disk from Alpha CD,  how to mount system properly?H Re: Backing up system disk from Alpha CD,  how to mount system properly?H Re: Backing up system disk from Alpha CD,  how to mount system properly? Re: BIND resolver library  Re: C file operations  Re: C file operations $ Re: HOw to convert seq to ascii file$ Re: HOw to convert seq to ascii file Re: Login using UAF? Help ' Long Distance shadowing options for VMS 1 Re: Massbus configuration issues and restrictions 1 Re: Massbus configuration issues and restrictions  Re: Mixed Cluster certification  Re: Off-the-wall CI Question Re: Off-the-wall CI Question Re: OT: Sun's fighting chance  Re: OT: Sun's fighting chance  Re: OT: Sun's fighting chance % Re: Sniffer Can't see cluster traffic % Re: Sniffer Can't see cluster traffic % Re: Sniffer Can't see cluster traffic % Re: Sniffer Can't see cluster traffic   Re: TCP/IP cluster interconnect? Re: user privilege( Re: VAX 6000 replacement with CHARON-VAX Re: VMS on IBM's Itanium Re: VMS on IBM's Itanium: Re: ZIP "-V" v. UNIX, et al.: Problem, possible solutions.: Re: ZIP "-V" v. UNIX, et al.: Problem, possible solutions.: Re: ZIP "-V" v. UNIX, et al.: Problem, possible solutions.< Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command< Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command< Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command< Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command< Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command< Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command< Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command  F ----------------------------------------------------------------------  # Date: Thu, 23 Sep 2004 18:41:59 GMT 4 From: "Fred Kleinsorge" <fred.nospam@nospam.dec.com>G Subject: Re: 1920x1200 resolution with VMS and the HP L2335 LCD monitor 3 Message-ID: <XNE4d.11568$pn7.3276@news.cpqcorp.net>   @ "Vance Haemmerle" <vance@toyvax.Glendale.CA.US> wrote in message4 news:GxD4d.1568$nj.631@newssvr13.news.prodigy.com... > Fred Kleinsorge wrote:D > > "Vance Haemmerle" <vance@toyvax.Glendale.CA.US> wrote in messageJ > >>   I look forward to see what you think.  Especially as I read there's4 > >>no VMS support for 1920x1200 on the Radeon 7500. > >> > >  > > L > > The Radeon driver table shows 1920 x 1200 at 60 and 70Hz in 8, 16 and 32 > > (24) bit depths. > 2 > Which VMS version(s)/patch(es) is this true for? >   < All.  I got the values directly from the driver.  Looking at4 DECW$DEVICE_CONFIG_GH.COM also lists these settings.  B > The most recent list of supported resolutions I could find, fromG > file RADEON_7500_README_VMS732.TXT in the VMS732_GRAPHICS-V0200 patch  > does not list 1920x1200... > L >                Table_2-1_Supported_Video_Modes____________________________ > 	 > Monitor = > Resolution________RADEON_7500_Card_________________________  > " > 640x480           60, 72, 75, 85 > " > 800x600           60, 72, 75, 85 > " > 1024x768          60, 70, 75, 85 > " > 1024x864          60, 70, 75, 85 >  > 1152x864          60 >  > 1280x1024         60, 75, 85 > " > 1600x1200         60, 65, 75, 85 >  > 1920x1440         60, 75L >                2048x1536_________60,_65,_70,_75___________________________ > . > The default video mode is 1024x768 at 70 Hz. >    Trust me, it's in there.   > E > >>   I guess my question is what is really the difference between a J > >>1920x1200 and 1600x1200 analog signal and is the P350s video differentK > >>between the two modes?  At a vertical refresh rate of 60Hz that means a K > >>horizontal rate of 72KHz, not counting the vertical interval size.  The L > >>monitor presets say 1600x1200 has a horizontal frequency of 75.00kHz butI > >>the 1920x1200 has 74.56kHz.  If it was a regular CRT it would stretch J > >>each line across the whole screen and wouldn't matter.  The LCD treats9 > >>the modes differently depending what it thinks it is.  > >> > >  > > A > > The P350 does 1920x1200 @ 24 bits at 60Hz with a dot clock of E > > 94.727479MHz  - while it has a 75Hz mode, this exceeds the RAMDAC 	 specs --  I > > so it isn't really supported at 75Hz reliably (the dot clock would be  > > 243.409349). > $ >    I guess you mean 194.727479MHz? > @ >    The monitor preset for this resolution and rate is 193.250. >   C Yes, I dropped the 1 in editing apparently (stupid PC keyboard ;-).   E > > 1600x1200 yields 161.795626MHz at 60Hz, and 202.500215MHz at 75Hz  > K >    This does closely match the monitor presets for these.  Pixel clock of ( > 162.00 MHz and 202.5 Mhz respectively. > > >    When I use 1920x1200 @60Hz, the monitor mode display says >  >    Resolution: 1600x1200 >    Horz Freq:  75.57 kHz >    Vert Freq:  60 Hz > J >    The Horizontal frequency of 75.57kHz  more closely matches the preset9 > of 75.00 for 1600x1200 than the 74.56kHz for 1920x1200.  > I > > The R7500 on the other hand has a dot clock of 193.16MHz at 1920x1200  @60Hz. > > K > > [video timings are strange things.  i think the p350 timings pre-date a  VESAG > > standard.  in fact, i think the radeon timings were from a proposed 
 > > standard]  > J >    Maybe that's the difference... the P350 isn't standard.  According toJ > the monitor specs, the stats for 1920x1200 is Horiz 74.56kHz, Vert 60Hz,I > pixel clock 193.25 MHz.  Which pretty closely matches the R7500 but not 3 > so closely the P350 (assuming you meant 194.727).  >   L That's my guess as well.  Generally speaking, the timings in use were pulled/ from a spec on the IBM RAMDAC used on the card.   = >    So maybe the L2335 monitor will correctly see the R7500.  >   + I'll let you know as soon as mine comes in.    > F > The file P300_350_V2_README.TXT doesn't say anything about not usingB > 75Hz and in fact says to use the highest value supported by your, > monitor (which is 60Hz for mine anyway)... >   E The code itself in this case shows it outside of the RAMDACs margins.      >  > ? >    I remember reading a post of yours a while back to someone = > who was asking what would be better, a P350 or an R7500.  I ? > think you stated that the P350 had better support of multiple > > pixel depth visuals and/or support for multiple simultaneous< > colormaps.  I remember when I set the default visual class? > to Pseudocolor it said that I would have to set the bit depth < > to 8.  xdpyinfo still said I had 24-bit visuals though.  I< > didn't have to set the bit depth to 8 on my ZLX-E2 to have: > PseudoColor and could have both 8-bit and 24-bit visuals7 > on the same screen.  Does the R7500 has this ability?  >   K No.  The ZLX-E2, like the P350, and some other workstation derived graphics J cards - used a RAMDAC originally designed by DEC, then BrookTree, and thenL IBM.  This expensive RAMDAC allocated the 8 bits in the high-order byte of aE 32-bit pixel to indicate how to interpret the pixel.  An even earlier L technology in the FireFox had a colormap window tree that was used to define  the pixel format for each pixel.  D This allowed a mixture of multiple depths on the screen and multiple
 colormaps.  A X11 exposed too much to applications in a way, such that too many 7 applications understand the pixel format on the screen.   K The cards that originated from the PC world (most of them these days) never G had that capability - Windows did need or use it.  So while they can be J programmed to a wide variety of modes - they generally don't have multipleK simultanious pixel formats, colormaps, or overlay planes.  Instead, you are 9 more likely to find video input (PIP) mixers for example.   K There isn't a 24-bit Pseudocolor mode (which means use the pixel value as a I colormap table lookup).  The pixel either generates TrueColor (each R/G/B K component drives a gun) or DirectColor (each R/G/B value is an index into a D table that gets the actual value to drive the gun).  You can emulate/ StaticGrey just forcing monotonic R/G/B values.   J PsuedoColor is only available with 8-bit pixels - it's an 8-bit index intoL an R/G/B table.  You can emulate TrueColor, and DirectColor (with very small  values for R/G/B) in 8-bit mode.  I So the default on the P350 and Radeon cards is 24-bit, TrueColor.  If you K change the depth, a default visual type will be applied.  If you change the I visual type default without changing depth - and the visual type makes no 1 sense at the current depth - you get the warning.    ------------------------------  % Date: Thu, 23 Sep 2004 20:34:48 +0200 - From: "Starlet731" <starlet731@attglobal.net> Q Subject: Re: Backing up system disk from Alpha CD,  how to mount system properly? 0 Message-ID: <roqdncHGb-mwis7cRVnygg@scarlet.biz>  > "Robert Young" <rcyoung@aliconsultants.com> schreef in bericht7 news:91437ce6.0409230440.6320e24d@posting.google.com... C > I am backing up the system disk on an Alpha to mag tape  by first . > booting from the install CD (7.3-1 OpenVMS). > D > However the instructions I have make it unclear whether the systemA > disk should be mounted "foreign" or not for the purposes of the E > backup.  Since I don't want to mess it up, can anyone fill me in on  > the proper setting?  >  > I was planning on using....  >  > init  mtu0:  sysbck ; > backup/image  source: mtu0:system.bck/rewind/label=sysbck   L You need to mount the system disk Files-11 in order to back it up correctly!  A If your system disk is called "DKA0" with label "LABEL" you type:    MOUNT DKA0: LABEL   H If you are restoring the system disk back from tape you have to mount it foreign:   MOUNT/FOREIGN DKA0: LABEL    ------------------------------  # Date: Thu, 23 Sep 2004 18:47:42 GMT # From: hoff@hp.nospam (Hoff Hoffman) Q Subject: Re: Backing up system disk from Alpha CD,  how to mount system properly? 3 Message-ID: <iTE4d.11570$Yl7.9708@news.cpqcorp.net>   o In article <91437ce6.0409230440.6320e24d@posting.google.com>, rcyoung@aliconsultants.com (Robert Young) writes: B :I am backing up the system disk on an Alpha to mag tape  by first- :booting from the install CD (7.3-1 OpenVMS).  ..C :However the instructions I have make it unclear whether the system @ :disk should be mounted "foreign" or not for the purposes of the
 :backup...  >   You're at the $$$ DCL prompt, so you have normal DCL access.<   If you mount the input volume /NOWRITE, you are reasonably   safe against command errors.   :I was planning on using.... :  :init  mtu0:  sysbck  @   No need to initialize the tape, as BACKUP can do that for you.  : :backup/image  source: mtu0:system.bck/rewind/label=sysbck  @   Something like the following (admittedly gonzo) BACKUP command   is what I typically use:  /   SET PROCESS/PRIV=ALL ! brute-force, obviously ,   MOUNT/FORIEGN mmcu:   ! target tape device:   MOUNT/OVERRIDE=IDENT/NOWRITE ddcu:  ! private disk mount,   BACKUP/IMAGE ddcu: mmcu:saveset.bck/save -A     /rewind/init/label=(x,y,z)/ignore=(label,access[,nobackup]) - 2     [/verify] [/noassist] [/record] [/block=32256]  >   There is a very useful list of sample BACKUP commands in the(   back of the BACKUP documentation, BTW.   --    N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com    ------------------------------    Date: 23 Sep 2004 14:33:24 -0700/ From: rcyoung@aliconsultants.com (Robert Young) Q Subject: Re: Backing up system disk from Alpha CD,  how to mount system properly? = Message-ID: <91437ce6.0409231333.17e62b83@posting.google.com>   F I appreciate the feedback. I will be away for a couple of weeks (henceA the backup), but will not have time to "test" it before the plane  leaves due to a "time crunch".    : > This implies that you need the disk mounted normally, so' > don't mount the disk /foreign. Either  >  > $ mount source label >  > or >   > $ mount source /override=label > J > The tape should be mounted /foreign. Backup will do this for you, but I K > prefer to issue a separate mount/foreign first as a double check to make  * > sure I'm not overwriting the wrong tape.   ------------------------------  + Date: Thu, 23 Sep 2004 18:51:09 +0000 (UTC) ? From: Graham Burley <burley.not-this@encompasserve-or-this.org> " Subject: Re: BIND resolver library9 Message-ID: <41531B00.243DFDCC@encompasserve-or-this.org>    JF Mezei wrote:  >   L > What a disaster, having spent/wasted so much time rolling my own because IP > couldn't find something on VMS, only to find that it is there but well hidden.   Did you consider NETLIB ?   ; It has a number of DNS routines including NETLIB_DNS_QUERY.    ------------------------------  # Date: Thu, 23 Sep 2004 18:35:37 GMT # From: hoff@hp.nospam (Hoff Hoffman)  Subject: Re: C file operations2 Message-ID: <ZHE4d.11566$Yl7.418@news.cpqcorp.net>  N In article <opser8ydfxzgicya@hyrrokkin>, "Tom Linden" <tom@kednos.com> writes:J :On Thu, 23 Sep 2004 13:15:21 GMT, Ed Vogel <edward.vogel_stop_the_spam.@= :hp.com> wrote:  :  :>@ :> "Chris Sharman" <chris.sharman@sorry.nospam> wrote in message1 :> news:cirsm6$3ag$1$830fa7a5@news.demon.co.uk... H :>> Is ther a way (supported or otherwise) to get at the underlying file$ :>> structures of standard C files ? :>> ; :>     I've never used this mechanism, but reading the CRTL @ :>     documentation, it looks like this is possible.  The fopen@ :>     function will accept all the optional arguments described= :>     in the creat function doc.  One of these optional args C :>     is called acc_callback.  It's followed by a function pointer > :>     that points to a call-back routine that is invoked just; :>     before the sys$create or sys$open call.  It's passed * :>     the address of the RMS FAB and RAB.    @   The acc extension works just fine -- I make regular use of it.A   You can read the FAB and RAB this way, but you should not alter (   the underlying FAB and RAB structures.  =   I'd tend to open the file shared if I needed both C and RMS ?   access to the file -- use the C channel for C operations, and %   the RMS channel for RMS operations.   A   And, um, what's a "standard C file"?  Part of the "fun" here is >   that C can easily access sequential, binary and even indexed9   files -- what problem(s) are you looking to solve here?     N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com    ------------------------------  % Date: Thu, 23 Sep 2004 15:38:44 -0400 ( From: David Froble <davef@tsoft-inc.com> Subject: Re: C file operations, Message-ID: <41532644.6050707@tsoft-inc.com>   Bob Koehler wrote:  i > In article <ciudt7$g86$1$8300dec7@news.demon.co.uk>, Chris Sharman <chris.sharman@sorry.nospam> writes:  > 8 >>Exactly what I want (the same as pas$fab/rab too ...). >> > J >    Yeah, and I wanted VAX Fortran style built in access to keyed indexedH >    files from the C RTL but nobody at DEC wanted to do that extension. >  >    Something like 3 >       kscanf(FILE*, keyvalue, format string, ...)  > : >    Or maybe a format specifier that denotes a key value. >  >   E If it was provided, for a fee, by a third party, would anyone buy it?    --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  # Date: Thu, 23 Sep 2004 18:57:27 GMT # From: hoff@hp.nospam (Hoff Hoffman) - Subject: Re: HOw to convert seq to ascii file 3 Message-ID: <r0F4d.11572$Yl7.8343@news.cpqcorp.net>   [ In article <4151A099.40305@ceris.purdue.edu>, Chuck Aaron <caaron@ceris.purdue.edu> writes: E :I have a sequential file of userid's. How can I convert the file to  5 :ascii to where when I send it it does not look like:  :  :useriduseriduseriduserid...  E   I'm going to make a whole series of assumptions here -- it's *very* A   helpful if details such as DIRECTORY/FULL and the mechanism for 8   sending the file and the target platform are involved.  F   There is no one single sequential file format, so what Windows mightF   call sequential doesn't necessarily match those formats that OpenVMS   might call sequential.    E   The most portable format for direct transfer to Windows is probably -   Stream, though StreamLF works quite well.     G   FTP transfers can and do work, assuming the correct mode is selected.   D   If the far end is OpenVMS and there are intermediate systems, thenC   I'd tend to package the file into a zip archive using current zip 2   and unzip tools for OpenVMS and the "-V" option.  F   If you need to convert a file from one format to another on OpenVMS,   the CONVERT command is used.  E   If you receive a file on OpenVMS that is stuffed up, you can try to H   use the SET FILE/ATTRIBUTES command to (probably) select either streamH   or stream LF (/ATTRIBUTES=STM or /ATTRIBUTES=STMLF) to reset the localD   attributes in the file header.  THIS DOES NOT CONVERT THE CONTENTSF   NOR THE RECORD STRUCTURE, so please don't assume that it does -- theD   SET FILE/ATTRIBUTES mechanism resets the file structure attributes    as stored in the file header.)  E   Depending on the input file structure and particularly on the input E   file record structures, you may well end up writing a tool -- using B   DCL or otherwise -- to read and convert the file to the requiredE   output file format for the transfer; you may have to write your own 6   EXPORT tool, and possibly even your own IMPORT tool.  F   Again, the OpenVMS version, the command sequence(s), the source and D   target platform details, the transfer command(s), etc., can all be   useful information, too.    N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com    ------------------------------  % Date: Thu, 23 Sep 2004 20:24:37 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>- Subject: Re: HOw to convert seq to ascii file + Message-ID: <41537754.7FFE8E57@comcast.net>    Chuck Aaron wrote: > E > I have a sequential file of userid's. How can I convert the file to 6 > ascii to where when I send it it does not look like: >  > useriduseriduseriduserid...  >  > but looks liks >  > userid > userid > userid  G As others have pointed out, that *IS* ASCII data. You can prove this to 1 yourself - just pull the file into a text editor.   G If the output you posted above emulates the result of the TYPE or PRINT  commands, see this URL:   - http://www.djesys.com/vms/mentor/rms.html#att    ...or...  ) http://www.djesys.com/vms/mentor/rms.html   E ...and read the entire page to learn about RMS files and about record  attributes.    D.J.D.   ------------------------------  % Date: Thu, 23 Sep 2004 20:19:45 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>" Subject: Re: Login using UAF? Help+ Message-ID: <41537631.A583180F@comcast.net>    RT Carter wrote: > A > I am a network tech but know nothing about VMS. Our VMS guy got H > "downsized" and I'm now taking care of three VMS servers. I need help.G > The workstations and servers were all on one IP network 192.255.1.xxx E > but we moved the designers to another building and the workstations F > are now 192.255.2.xxx (servers stayed the same). I know that we haveF > security set to only allow certain IP addresses into the VMS serversC > but I don't know where to change the access from 192.255.1.xxx to F > 192.255.2.xxx. It isn't in the login.com files. I've read about UAF.G > Is this were the IP address would be defined? Where are the UAF files D > kept--is there a standard name for the files? Thanks for any help.  D Other respondents have made reference, but have not fully explained:  C TCP/IP is essentially "foreign" to VMS. In fact, VMS will run quite % happily with no network stack at all.   ( There are three TCP/IP products for VMS:   TCPware and Multinet: 1 ...from Process Software, http://www.process.com/   6 TCP/IP Services for OpenVMS (formerly known as "UCX"):E ...from the company formerly known as Digital(, then Compaq, now hp).   G Another poster has suggested how to determine which one you are dealing  with.   H Others have also suggested that the UAF knows nothing of TCP/IP. Indeed,? VMS itself knows nothing of TCP/IP - it's an add-on, as we just 
 discussed.  @ You will likely be looking into the network to determine how the@ restrictions wre imposed before you need look at the VMS system.  
 Good luck!   D.J.D.   ------------------------------  % Date: Fri, 24 Sep 2004 12:46:18 +1000 , From: "mjw" <wademjx@nospam-optusnet.com.au>0 Subject: Long Distance shadowing options for VMS< Message-ID: <41538a83$0$22860$afc38c87@news.optusnet.com.au>  F In my environment we have a number of Multisite OpenVMS V7.3-1 and -2 M clusters using dual Fibre Channel links for IO and dual Ethernet for cluster  F comms.  This is all fine, works well and the performance is more than ) acceptable over an approx. 10km distance.   K My next trick will be to replicate this cross site shadowed data over long  C distance (>1000km) for true DR.  Now obviously this has to be done  J asynchronously to avoid delays which is fine but how to do this?  I still K want to maintain my local site to site shadows using VMS shadowing to give   me site disaster tolerance.   4 My goal is to have a start of day position in place.  H I have thought about EVA to EVA Continuos Access but the configurations K required, minimum dual link requirements etc etc are starting to look very  K unactractive from a cost perspective especially give my normal transaction   rates.  F Anyone know of any third party options out there or do I just look at M shipping daily incremental backup copies which might have issues due to time   constraints and data volumes.    Thanks in advance, Malcolm    ------------------------------  % Date: Thu, 23 Sep 2004 22:57:08 -0400 , From: Bob Supnik <bob.supnik@sun.nospam.com>: Subject: Re: Massbus configuration issues and restrictions8 Message-ID: <t437l09bb6mus2cdu3l04cdukh1dtqdc51@4ax.com>  . This RP/RM Massbus stuff gets worse and worse.  @ I was working on the assumption that RP's and RM's had differentD offsets for SN (= 12 on RP and 8 on RM) and ER2/MR2 (=8 on RP and 12 on RM) because    1) VMS' drivers show it that wayD 2) Published RP/RM documentation (including the maintenance manual!) shows it that way    But, digging deeper   ? 1) TOPS-10 and TOPS-20 drivers show the RP SN = 8 and ER2 = 12, E 2) PDP-11's clearly supported mixed strings, which would only work if  RP SN = 8 and ER2 = 12, and B 3) RH11/RH70 documentation shows only one register map, which only works if RP SN = 8 and ER2 = 12 ? 4) Ultrix relies on SN being at 8 regardless of the drive type.    And the capper  : 5) The RP04/05/06 <schematics> show RP SN = 8 and ER2 = 12  : The RP04/05/06 does the register decodes with a 74154 4:16? demultiplexor.  The selects are laid out in numeric order, with   
 8 = SN and	 12 = ER2.   F So, the weight of the evidence is that, in fact, the RP SN = 8 and ER2E = 12.  This means the RH11/RH70 had only a single PROM map, which was ? consistent for RP, RM, and TU; and that mixed strings worked on B PDP-11's and PDP-10's; and that the technical manual is incorrect.  2 Then what is going on with VMS???  Two hypotheses:  F 1) RP06's were 'different' on VAXen.  Seems unlikely, unless VAXen had> a different board set in the disk.  The decoder is not a PROM,, interchanging 8 and 12 would be an etch cut.D 2) VMS doesn't care.  Even though SN and ER2 are defined, the driverC (niether the mainline driver nor the boot driver) doesn't use them. B If they are inverted, VMS would not notice.  If VMS adhered to the? published documentation, it would have the 'wrong' definitions.   F But VAX diagnostics <would> care.  Anyone have VAX diagnostics for theE RP04/05/06?  Or a real VAX with real RP06's still running, from which  I could get a register dump?  E The 'rule' of primary sources says that the schematics are right, and E that deviates from the schematics -- including the maintenance manual F and the VMS driver -- is wrong.  I'd sure like someone to come up with> an alternate explanation that encompasses all the known facts.   /Bob Supnik   . On Mon, 20 Sep 2004 21:55:10 -0400, Bob Supnik" <bob.supnik@sun.nospam.com> wrote:  E >In preparing to simulate the VAX 11/780 (in hopes of bringing up VMS G >V1), I dug more deeply into the workings of the Massbus.  I discovered D >a rather odd problem: the RP and RM series of drives are not reallyF >compatible. The issue is that internal registers 8 and 12 are not the >same  >  >reg		RP		RM >  >8		ER2		SN  >12		SN		MR2 > C >In the VAX, the Massbus adapter doesn't hide the internal register F >numbering of the disk controllers, and this discrepancy is visible toG >software.  As a result, VMS can handle 'mixed' RP and RM drives on the  >same Massbus adapter. > G >But on the -11 (and on the KS10), SN is always 176730, for both RP and E >RM drives.  This implies that 176730 is being mapped differently, in F >the RH11/RH70, for the different drive types.  Also, on the RH70, BAEE >and CS3 appear at different locations for the TU than for the RP/RM.  > D >This is all explained by the fact that the RH adapter had a mapping@ >PROM that mapped Unibus addresses to internal/external registerD >numbers.  The PROM was clearly different for an RP, an RM, or a TU.C >So what happened when RP's and RM's were mixed on the same Massbus E >adapter???  (Disks and tapes were never mixed.)  If the RH had an RP F >map, addresses 30 and 40 would return the wrong values for RM drives;6 >in an RM map, 30 and 40 would be wrong for RP drives. > E >TOPS-10 has a single driver for "RH" drives.  It reads SN and stores D >it in the UDB; that's never referenced by the driver again.  ER2 isB >only zeroed and never read.  MR2 is never referenced.  So TOPS-10 >should run with mixed strings.  > E >RSTS/E and RSX have separate drivers for DB (RP) and DR (RM) drives. E >RSTS/E will catalog a Massbus and assign the units to DB and DR.  In F >RSTS/E, the DB (RP) driver never reads the SN (it's not even defined)G >or MR2, and it bypasses the read of RPER2 if the drive is an RM, so it  >too looks to be ok. > F >RSX also doesn't define the SN.  But the DB driver will look at RPER2E >in attempting to do an ECC correction.  Under an RM map, it will get F >the wrong register from the drive for ER2. The DR (RM) driver doesn't2 >look at either SN or MR2 and thus can't go wrong. >  >So here are my questions: > E >1) Is my understanding of the RH11/RH70 correct, that is, were there 0 >different address mappings for RP vs RM drives?B >2) Were mixed strings (RP's + RM's on the same Massbus) supported >under TOPS-10/TOPS-20? B >3) Were mixed strings (RP's + RM's on the same Massbus) supported >under RSTS/E? under RSX?  >  >/Bob Supnik   ------------------------------  % Date: Thu, 23 Sep 2004 20:13:19 -0700 + From: Mark Crispin <mrc@CAC.Washington.EDU> : Subject: Re: Massbus configuration issues and restrictionsI Message-ID: <Pine.LNX.4.62.0409232005570.28196@shiva0.cac.washington.edu>   & On Thu, 23 Sep 2004, Bob Supnik wrote:4 > Then what is going on with VMS???  Two hypotheses:& > 1) RP06's were 'different' on VAXen.  K Possible.  Some disk Massbus controllers did have to be rewired if used on  J a different processor.  I remember 18-bit mode as being one of the issues.  " > Seems unlikely, unless VAXen had@ > a different board set in the disk.  The decoder is not a PROM,. > interchanging 8 and 12 would be an etch cut.   Not a wirewrap cut?    > 2) VMS doesn't care.  # Very likely.  TOPS-20 doesn't care.   G > The 'rule' of primary sources says that the schematics are right, and G > that deviates from the schematics -- including the maintenance manual ! > and the VMS driver -- is wrong.   I This is probably correct.  My guess is that it was a typo that never got   fixed because it didn't matter.   
 -- Mark --   http://staff.washington.edu/mrc F Science does not emerge from voting, party politics, or public debate. Si vis pacem, para bellum.   ------------------------------  % Date: Thu, 23 Sep 2004 15:25:55 -0400 * From: Joel Loveless <joell@mindspring.com>( Subject: Re: Mixed Cluster certification8 Message-ID: <ri86l051r2a54stqgomj8c6ppir81d6toh@4ax.com>  F Here is a link to the VMS 7.3-2 release notes. It indicates in sectionD 1.4 that going from 7.2-1 directly to 7.3-2 is supported on a singleD systems. Just below that, it indicates that running 7.2-1 with 7.3-2@ in a mixed cluster is not supported. It list the supported mixed versions. Hope this helps.  8 http://h71000.www7.hp.com/doc/732FINAL/6668/6668PRO.HTML   ------------------------------  % Date: Thu, 23 Sep 2004 20:10:15 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>% Subject: Re: Off-the-wall CI Question + Message-ID: <415373F7.E8DD5532@comcast.net>    Bob Willard wrote: > [snip]J > Look for a cheap star coupler.  It supplies the right signal attenuationL > and isolation and, as VB reminded me, is needed to make the CIPCA pass the > loopback selftest.  G AH! So, I gather (in passive mode) that the Star Coupler is bigger part H of the design than simply a "star coupler": it has electrical properties that are expected, as well.   	 So be it.    D.J.D.   ------------------------------  % Date: Thu, 23 Sep 2004 20:08:10 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>% Subject: Re: Off-the-wall CI Question + Message-ID: <4153737A.958D0841@comcast.net>    Marty O'Connor wrote:  > A > "David J Dachtera" <djesys.nospam@comcast.net> wrote in message ' > news:4150DCC1.E8ECA795@comcast.net...  > : Keith Parris wrote:  > : >  > : > [snip]C > : > I'm sure this is freeing up CI hardware all over the country.  > : J > : Not likely. Existing clusters tend to stay that way (the SysAdmins gotI > : laid-off years ago, and they've been running without a reboot since - 5 > : clueless Borg drones don't like to mess with it).  > :  >  > Where are you located?  > Suburban Chicago. (Gotta put the website URL back in my .sig!)   D.J.D.   ------------------------------  % Date: Thu, 23 Sep 2004 15:33:16 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> & Subject: Re: OT: Sun's fighting chance, Message-ID: <415324FC.9DE61861@teksavvy.com>  3 You know, this talk about "open" got me thinking...   J Imagine IF, unhindered by the unwanted port to IA64, the VMS engineers hadH managed to give VMS a linux compatibility layer, complete with FX32! andE delivered this FASTER than Sun's equivalent which is expected "soon".   = The marketing opportunities would have been very interesting.   D VMS could run the shrinkwarpped Linux software are "OK" speed, or byM recomopiling the open sources linux applications, you'd get full alpha speed.    ------------------------------  % Date: Thu, 23 Sep 2004 15:29:34 -0400 ( From: David Froble <davef@tsoft-inc.com>& Subject: Re: OT: Sun's fighting chance, Message-ID: <4153241E.8030309@tsoft-inc.com>   Bob Koehler wrote:  g > In article <Xns956D5859BFEC6dcovmsrox@212.100.160.123>, "Doc." <doc.cypher@openvms-rocks.com> writes:  >  > K >>Why am I not surprised that you don't understand what open source is Bob?  >>J >>The "Open" in OpenVMS is to indicate that it complies with certain open  >>standards. >> > G >    To me "open" means I can look inside.  At the current price of the ? >    CD I can, if I need to.  For Linux I can.  For MS I can't.  >   >    Is Solaris open source now? >  >   O What's really humorous is that you can ask 10 people what 'open source' means,  $ and you'll get 10 different answers.  ( To some, it means free/in public domain.   To some, it means standards.  ( To some, it means access to source code.   .  .  .   % The ensuing arguments can be comical.   R My solution, say exactly what you mean, in clear terms, and avoid the wasted time.   Dave   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Roadh Vanderbilt, PA  15486e   ------------------------------  % Date: Thu, 23 Sep 2004 15:35:02 -0400l( From: David Froble <davef@tsoft-inc.com>& Subject: Re: OT: Sun's fighting chance, Message-ID: <41532566.2000909@tsoft-inc.com>  3 Andrew (Andy Boy) Harrison SUNUK Consultancy wrote:i   > Bob Ceculski wrote:t > C >> Andrew Harrison <andrew_remove__s._harrison@s_u_n.com> wrote in p7 >> message news:<cis6vr$dj2$1@new-usenet.uk.sun.com>...r >>F >>> Sun has no intention of attacking NetBSD or FreeBSD, RedHat is theE >>> target not them and OpenVMS despite having the Open bit tacked onu# >>> the front is hardly OpenSource.  >>>  >>> Regardsh >>> Andrew Harrison  >> >> >>@ >> it certainly is open source ... you can port any c unix/linux; >> garbage code onto it including libraries and it runs ...i= >> and you can get the source if you want to ... but vms doesi< >> not need to be tinkered with by people who think they can2 >> code understand os's but don't (i.e. linux) ... >  >  > 4 > I suggest that you broadcast your re-definition of9 > OpenSource to one of the Linux newsgoups, the responsesu  > will make interesting reading.  Q I was going to open with "Who wants to waste time with what they think?", but as SG I really have no idea what they think that would be a bit presumptuous.e  J My impression is that they think software should be free.  If so, I still T haven't figured out how they pay the mortgage.  Or is Linux the OS for the homeless?   Dave   -- a4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road- Vanderbilt, PA  154862   ------------------------------  # Date: Thu, 23 Sep 2004 18:06:04 GMT./ From: "Jeff Goodwin" <jgoodwin@maine.rrr-r.com>a. Subject: Re: Sniffer Can't see cluster traffic5 Message-ID: <ggE4d.8912$yg.5496@twister.nyroc.rr.com>   : "James O'Shea" <seamas_ose@ameritech.net> wrote in message: news:20040922151612.64562.qmail@web81107.mail.yahoo.com...@ > We have set up a small cluster to monitor cluster traffic. TheL configuration consists of:  Two clustered MicroVAxen, V6.2 VMS, plugged intoJ a hub with a sniffer monitoring traffic. MicroVaxen are MSCP serving disks2 to one another and disks are mounted cluster-wide. > L > Our network guy, who has never monitored a VMS cluster previously, says he is only see multicast packets  >. > G > The LAVC traffic protocol he's seeing uses a multicast MAC address toGJ communicate with all workstations in the broadcast domain. AB0004010EF7 is for the cluster in the Lab >iJ > The first part of the MAC address classifies it a DEC LAVC frame and the6 ethertype of 6007 also identifies it for the protocol. >o: > AB-00-04-01-xx-yy 6007 DEC Local Area VAX Cluster groups > ' > Sys. Communication Architecture (SCA)_ >_ >_ >_E > In his tests he's  only seeing traffic to the multicast address. In G contacting HP/VMS support they express that the described  LAVC traffic H should be expected   but also said we should see some MAC to MAC trafficI between the two MicroVaxen systems. So far we do not see the two machines J direct packets to any MAC address other than the Muticast MAC address. TheK multicast frames only are a 128bytes large and seem to have a window of one  (36 bytes appears to be data). >sK > We've done copies and backups from one node to the other but he says he's6E till not seeing traffic other than multicast packets.  Should this be 	 expected?  >a	 > Thanks,e > Jim O'Shea >z >a   ------------------------------  # Date: Thu, 23 Sep 2004 18:08:12 GMTv/ From: "Jeff Goodwin" <jgoodwin@maine.rrr-r.com> . Subject: Re: Sniffer Can't see cluster traffic5 Message-ID: <giE4d.8913$yg.8608@twister.nyroc.rr.com>   : "James O'Shea" <seamas_ose@ameritech.net> wrote in message: news:20040922151612.64562.qmail@web81107.mail.yahoo.com...@ > We have set up a small cluster to monitor cluster traffic. TheL configuration consists of:  Two clustered MicroVAxen, V6.2 VMS, plugged intoJ a hub with a sniffer monitoring traffic. MicroVaxen are MSCP serving disks2 to one another and disks are mounted cluster-wide. > L > Our network guy, who has never monitored a VMS cluster previously, says he is only see multicast packetsn >a >i  H Others have mentioned this, but perhaps not clearly.  You're most likelyK hooked up to a network device that acts as a switch, not a repeater.  If ittF is a switch, you will only see packets whose destination/source is theH sniffer.  This would also include multicast and broadcast packets.  MostL network gear is now switch gear.  We keep a few old repeaters around just to  monitor devices on switch ports.  L It's possible that the network device you have supports port mirroring, thusD you could mirror one of the VAX ports to the port which connects the sniffer.   -Jeff   + (sorry about previous non-value added post)    ------------------------------  % Date: Thu, 23 Sep 2004 15:57:15 -0400i- From: JF Mezei <jfmezei.spamnot@teksavvy.com>A. Subject: Re: Sniffer Can't see cluster traffic, Message-ID: <41532A99.581AE568@teksavvy.com>   James O'Shea wrote: 8 > Two MicroVaxen running VMS v6.2 are connected, via cat5 > 5, to a Cisco smart hub.  Nothing else is connectedk4 > except his laptop which is running Network General
 > sniffer.  G The word "smart" in "smart hub" is the big question mark here. They are) normally incompatible.  N You should find the documentation for your exact Cisco device and look for any4 information on enabling promiscusous mode on a port.  G The fact that you are seeing multi and broadcasts from the cluster does N indicate that you have not setup a VLAN that would totally isolate the cluster" traffic inside a group of 2 ports.  I Another way to look at this is to inidiate a large data transfer from onerM machine to another (for instance, DIR/FULL from node 1 of a whole disk servedMN by NODE 2). While this is going on, look at the actual lights on the Cisco boxK ports. You should see plenty of flashing between the 2 ports connecting the M vaxes, and a much reduced flashing on the port used by the sniffer would also K be an indication that the Cisco device isn't giving you all of the traffic.e   ------------------------------  + Date: Thu, 23 Sep 2004 20:39:41 +0000 (UTC)e7 From: moroney@world.std.spaamtrap.com (Michael Moroney)i. Subject: Re: Sniffer Can't see cluster traffic( Message-ID: <civcad$qch$1@pcls4.std.com>  / "John E. Malmberg" <wb8tyw@qsl.network> writes:s   >JF Mezei wrote: >> "John E. Malmberg" wrote: >> yN >>>The SCS protocol is non-routable, so it should show up on all switch ports. >>   >> Can you explain this ?g  H >As I remember from looking at the protocol traces, they were sent to a J >protocol and cluster specific multicast MAC address, from a protocol and " >node specific source MAC address.  G A protocol such as SCS uses "hello" packets sent to a multicast addresssJ (that all nodes using the protocol listen to) to find each other.  PacketsE directed to a specific node (and only the specific node) use the MAC oD address of the node's ethernet adapter, so other nodes won't see it.E (because it was either filtered by other node's ethernet adapter, or r& nowadays, not passed on by the switch)  H >There are bits in the packets that the switches use to determine if it - >is a multi-cast packet or a directed packet.   H Yes.  The multicast bit is set in the "hello" packets and not set in the directed data packets.  F >Typically the characteristics of a non-routable protocol are that it I >communicates with broadcast destination addresses so it will not depend  G >on keeping a mapping table of MAC addresses to the remote hosts.  The /H >upper level of the protocol handler then filters out what packets that  >it cares about.  G Wrong.  SCS and LAT do keep MAC address tables so that directed packetssH are only seen by the targeted node.  Why make other nodes keep throwing " away packets not directed at them?  G IP uses the "broadcast address" of FF-FF-FF-FF-FF-FF as its own versionsF of hello packets for finding each other on a LAN (it actually uses ARPH for this).  As far as I know there was never a real definition as to howE the broadcast address was to be used, so several protocols other thaniB IP/ARP also use it.  Digital-derived protocols such as SCS and LATE never use the broadcast address, using a protocol-specific multicast  D address instead.  This way, nodes not interested in, say, SCS, never> see any SCS traffic because they don't listen to the multicastA AB-00-04-01-xx-yy hellos or directed packets (which will never be7 directed to them).  G Because of how switches operate, a node trying to monitor SCS will only4E see the multicast hellos and not the node-to-node traffic, unless the * monitoring port has been made promiscuous. --   -Mike    ------------------------------  % Date: Thu, 23 Sep 2004 19:39:30 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>) Subject: Re: TCP/IP cluster interconnect?o+ Message-ID: <41536CC2.73652D6C@comcast.net>    JF Mezei wrote:i >  > David J Dachtera wrote:iM > > > > The point is that you'd end up doing a lot of "front end" work (stuffi/ > > > > that gets set up before VMS even boots)  > >o( > > Is that not what you just described? > L > Not sure. The way I had read the above was similar to SCS which gets setup' > very early in thge VMS boot sequence.    > David J Dachtera wrote:sM > > > > The point is that you'd end up doing a lot of "front end" work (stuff / > > > > that gets set up before VMS even boots)   2 What part of "before VMS even boots" is ambiguous?  J > What I described was a puerly hardware based solution with a TCPUP stackP > implemented in a hardware card which becomes available as soon as the power is% > applied (eg: even in console mode).s  + Read me one more time and you might get it.o  7 (Hint: Remember the possible values for "AUTO_ACTION".)n   D.J.D.   ------------------------------  # Date: Thu, 23 Sep 2004 19:10:44 GMTH# From: hoff@hp.nospam (Hoff Hoffman)t Subject: Re: user privilege 4 Message-ID: <UcF4d.11574$Yl7.11018@news.cpqcorp.net>  ^ In article <de74637.0409211150.436d965b@posting.google.com>, cljlk@hotmail.com (cljlk) writes: :I am not a VMS sytem person.E9 :Would anyone please show me how to check user privilege?w5 :I know "show user" for checking users on the system. ! :How to check the user privilege?B% :And how to set privilege for a user?m  C   Please take the time to skim the OpenVMS User's Guide as a start,eA   and then -- if you are to manage OpenVMS systems -- please then >   skim through the OpenVMS System Manager's Essentials Manual.  >   Folks here can answer direct questions, but we may well miss@   critical considerations for your particular environment -- the?   OpenVMS manuals try to provide you with both the information [<   for the particular option or operation, as well as related;   considerations and details you might not have thought of.d  C   If you need more formal assistance or training, there are classes-F   and materials available -- the OpenVMS FAQ has pointers -- and thereE   are HP and third-party organizations that can provide you with thisa@   training, with support, and with system management assistance.  N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com7   ------------------------------  % Date: Thu, 23 Sep 2004 19:36:25 -0500I2 From: David J Dachtera <djesys.nospam@comcast.net>1 Subject: Re: VAX 6000 replacement with CHARON-VAX + Message-ID: <41536C09.443814D8@comcast.net>    Robert Boers wrote:e > A > "David J Dachtera" <djesys.nospam@comcast.net> wrote in messagee' > news:4150CE73.416A2416@comcast.net...  > K > > What do you have up your sleeve for data migration? (DSSI/CI storage toB > > DAS/iSCSI/whatever?) > > 
 > > D.J.D. > L > Standard VMS tape backup procedures. We modified the KDM70 disk controllerN > in the VAX 6000 emulator to support fast SCSI tape drives for data exchange.  H I trust that includes "fast SCSI" models of TZ8x drives such as TZ87 and TZ88...?  I > Supporting existing DSSI or CI storage hardware is a non-goal, as theseDG > components are comparatively slow, have a low storage density and aree > expensive to maintain.  D However, a simple plug-and-play replacement for a VAX is going to beG easier for your resellers to push than a full replacement of the entire G computer/storage system. See, people keep their VAXes because they just-# work. Likewise, their storage farm.,  D When processing speed becomes an issue, they look for a step-up fromD their VAX. When (If?) storage capacity and I/O throughput becomes anE issue, THEN - and ONLY then - will they go shopping for replacements.d  / > Dedicated iSCSI targets have often integrated,L > redundancy and a very high storage density. Maintenance cost and footprint= > reduction are important factors in switching to CHARON-VAX.:  
 Well, o.k.  @ ...but the question was about immediate VAX replacement and data" migration, not long term upgrades.   D.J.D.   ------------------------------  % Date: Thu, 23 Sep 2004 20:03:29 -0500t2 From: David J Dachtera <djesys.nospam@comcast.net>! Subject: Re: VMS on IBM's Itanium>+ Message-ID: <41537261.468F4DC6@comcast.net>g   Bob Ceculski wrote:  > c > JF Mezei <jfmezei.spamnot@teksavvy.com> wrote in message news:<4151D461.3705CC99@teksavvy.com>...y > > Bob Ceculski wrote: K > > > Your recent e-mail query to Sam Palmisano regarding an article in the D > > > Inquirer.net has been forwarded to me for review and response. > >AG > > > most enterprise customers are now looking for more open operating I > > > environments. As noted in the article, Compaq, and now HP, have not L > > > invested to keep up with the requirements of the customer base and the8 > > > result is an inevitable decline in installed base. > > R > > Well, bob, I think you did a great job in contacting IBM, even if the responseR > > was brutal and true. Just make sure nobody at HP sees it because they will use3 > > it as further justification to deemphasize VMS.  > H > what truth?  She states that there are several attractive alternatives > to vms ... what are they?   F Well, remember who we're discussing. "Attractive" is likely more to do& with appearances than business acumen.   D.J.D.   ------------------------------  % Date: Thu, 23 Sep 2004 22:10:34 -0400)# From: "John Smith" <a@nonymous.com>[! Subject: Re: VMS on IBM's Itanium-, Message-ID: <55-dnWAidouHH87cRVn-vQ@igs.net>   David J Dachtera wrote:a > Bob Ceculski wrote:j >>; >> JF Mezei <jfmezei.spamnot@teksavvy.com> wrote in messagen+ >> news:<4151D461.3705CC99@teksavvy.com>...< >>> Bob Ceculski wrote:sF >>>> Your recent e-mail query to Sam Palmisano regarding an article inG >>>> the Inquirer.net has been forwarded to me for review and response.a >>>dF >>>> most enterprise customers are now looking for more open operatingH >>>> environments. As noted in the article, Compaq, and now HP, have notG >>>> invested to keep up with the requirements of the customer base and ; >>>> the result is an inevitable decline in installed base.  >>> E >>> Well, bob, I think you did a great job in contacting IBM, even if F >>> the response was brutal and true. Just make sure nobody at HP seesG >>> it because they will use it as further justification to deemphasize  >>> VMS. >>< >> what truth?  She states that there are several attractive >> alternatives- >> to vms ... what are they? >2H > Well, remember who we're discussing. "Attractive" is likely more to do( > with appearances than business acumen.     Attractive:  As in,84 1) items which customers find palatable to purchase;I 2) usually preceeded by advertising and marketing to influence perception  and purchasing decisions;.I 3) items which are not deemed to be in risk of EOL the day after they aret purchased - see 2) above.e   ------------------------------  + Date: Thu, 23 Sep 2004 16:30:49 -0500 (CDT)t From: sms@antinode.orgC Subject: Re: ZIP "-V" v. UNIX, et al.: Problem, possible solutions. ) Message-ID: <04092316304915@antinode.org>   %    Today's update for the fascinated:e  + From: shoppa@trailing-edge.com (Tim Shoppa)a  Date: 20 Sep 2004 10:53:09 -0700  H > "-V" zip archives unpack exactly on real VMS systems.  (Or, well, they0 > did, until you made your modification.)  [...]  G    Actually, they didn't.  When the Zip 2.3 code was reading a file forwF a -V archive, it quit too soon to get all the allocated blocks, exceptG in a few special cases (disk cluster size = N* 32).  Of course, in mostiH cases, some accurate post-EOF data are read, and in many cases the extra5 data will all be zeros, so this might not be obvious.r  E    My early observations showed me that the current code was far from E reading all allocated blocks, and that led me to conclude that it was H not intended to do it.  Since then, I've consulted with some experts whoH carry the oral tradition (which was particularly helpful, as the writtenD record is not very informative).  They have convinced me that it wasE actually intended to do this, and, in fact, it's not hard to arrange.   E    One lingering question follows.  The upper-level Zip code tries tosH read 64KB chunks.  The current VMS (-V) code (QIOW, IO$_READVBLK) breaksG this down, and attempts to read 16KB chunks, while there seems to be no G obvious reason not to read 32KB chunks.  Does anyone know of a disk (oriC other file-structured?) device where a 32KB read may be expected tovF fail?  (Assume that there are at least that many bytes available to be read.)  F    For the morbidly curious, my latest changed files are available at:  +       http://www.antinode.org/ftp/info-zip/)&       ftp://ftp.antinode.org/info-zip/  H ------------------------------------------------------------------------  4    Steven M. Schweda               (+1) 651-699-98183    382 South Warwick Street        sms@antinode-org-    Saint Paul  MN  55105-2547i   ------------------------------  % Date: Thu, 23 Sep 2004 19:56:09 -0500g2 From: David J Dachtera <djesys.nospam@comcast.net>C Subject: Re: ZIP "-V" v. UNIX, et al.: Problem, possible solutions.h+ Message-ID: <415370A9.C10B61E3@comcast.net>:   Soterro wrote: >  > David J Dachtera wrote:4K > > Learn to use ZIP and UNZIP to your advantage instead of working againstaJ > > yourself. Then, you can scrap the perl stuff or keep it for reference. > I > I find quite difficult to use this version of ZIP to my advantage, whenoE > my advantage is to create the right files on VMS but also usable ont5 > Solaris (and readable on Windows). I simply cannot.f  E Go through the presentation again(, and again, and ...). Download theoG .PPT if that's a better medium for you. View/Print it in the Notes view  so it reads like a textbook.  A ZIP has the options you need. Learn to use them. See the slides IhF referenced in the earlier post, and the results of using those options8 in another sub-thread (supporting Steve's observations).   > I felt free to hackeJ > it to the best of my knowledge with those 'scrappable' Perl scripts, andF > it worked as it should have been. Tu my bemusement I see that a realK > solution even exists but is rejected from the start as 'utter silliness'.t   By whom?  E Make as much work for yourself as you like. I'm just telling you that G the work has already been done. All that remains is to learn how to use. the product.  B > So what if it works as documented? Change then the documentation  E Well, sorry to have to report that Phil Katz died a couple years ago. H So, changing the specification ("documentation") is not an option unless4 someone else wants to take ownership of the product.   > first,+ > people are not living by the Book Of Law.l  3 Yes. We see that - in Iraq, Afghanistan, Sudan, ...    > VMS was forced to adopth	 > TCP/IP,   C "Forced" is perhaps the not the right word. Given that freeware and G third-party TCP/IP stacks for OpenVMS existed before UCX, it was likely  a competitive choice.f  
 > run Apache,n   ...competitive choice.   > port Mozilla,e   competitive choice.    > take out PostScripto  . Bad decision by a vendor beyond VMS's control.   > and so on, was > ever evolution a bad thing?P  
 Read much?  & > Incidentally it was pressured by the1 > customers demand, not by developers decision :)a  A Well, yes and no. It was pressured by customers' CHOICEs, and thenB competitive response was to make certain choices to match customer needs.   > Crossplatform is all$ > over, and ZIP will have to follow.   It does. Learn how to use it.c  % > This of course only if VMS _future_  > is of any interest.e  H It is. However, remember that the needs of the installed base far exceedC the need to keep up with bg, Inc. or any of the UN*X-land stuff out, there.  F > Please don't get me wrong, the way all you community guys handle anyJ > help pleas is remarkable, kudos for that. I don't mean anything bad with7 > my note. The problem is not the past, but the future.   G It may be more accurate to say that the past, combined with the presentnG will point the way to the future; until then, however, the needs of thee installed base take precedence.i   > Maybe a look intoeE > the way Sun is handling (some) ideas from the openworld could bringtD > something. Don't blame for _all_ the problems the HP management :P > 5 > Here were another 2 cents, take them or leave them.   D Well, I'll take them, and try to turn them into something for you to; learn. The choice to learn or not is, of course, up to you.   ? > PS: Sorry that I used HP and management in the same phrase :D    We all make misteaks.a   D.J.D.   ------------------------------  % Date: Thu, 23 Sep 2004 20:01:05 -0500u2 From: David J Dachtera <djesys.nospam@comcast.net>C Subject: Re: ZIP "-V" v. UNIX, et al.: Problem, possible solutions.m* Message-ID: <415371D0.12F5B8C@comcast.net>   sms@antinode.org wrote:t > ' >    Today's update for the fascinated:c > - > From: shoppa@trailing-edge.com (Tim Shoppa)t" > Date: 20 Sep 2004 10:53:09 -0700 > J > > "-V" zip archives unpack exactly on real VMS systems.  (Or, well, they2 > > did, until you made your modification.)  [...] > I >    Actually, they didn't.  When the Zip 2.3 code was reading a file foriH > a -V archive, it quit too soon to get all the allocated blocks, exceptI > in a few special cases (disk cluster size = N* 32).  Of course, in most J > cases, some accurate post-EOF data are read, and in many cases the extra7 > data will all be zeros, so this might not be obvious.v  > So, if I follow this correctly, both ZIP and UNZIP need a fix?   ...not just ZIP?   D.J.D.   ------------------------------  % Date: Thu, 23 Sep 2004 15:23:03 -0400o( From: David Froble <davef@tsoft-inc.com>E Subject: Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command , Message-ID: <41532297.1030308@tsoft-inc.com>   Doc. wrote:   D > %NEWS-I-NEWMSG, Charlie Hammond wrote in news:r5A4d.11515$hV6.8250 > @news.cpqcorp.net  >  > C >>Although I've DIF'd a lot of DCL procedures, I am having trouble t; >>understanding the problem here.  Did I miss and example? e& >>If so, perhaps you would re-post it. >>7 >>I have found that /IGNORE=SPACING occasionally helps.  >>Have you tried that? >> > F > Alan's original post suggested the problem lies with lines that are J > comments.  As I understand it he said that DIFFERENCES with the various K > flags to ignore comments treated a line "$! This is a comment" as if the ;J > line actually said "$" and could get itself confused and start matching J > these lines with similarly treated comments at a different point in the 
 > other file.n > . > I believe the behaviour he's seeking is thatH > 1)  Lines with ! comment delimiters are stripped back to "$". (normal)1 > 2)  Lines have the starting "$" stripped. (new) J > 3)  Comment lines that started "$!" are now blank and treated as thus by >     DIFFERENCES. >  >  > Doc. >   P I know it's not part of the question/suggestion, but some of us, maybe just me, Q think comment lines are important.  If they are different, it might be important eK to see the differences.  I have this quaint notion that any code should be o< managable.  Out of date or errorneous comments can bite you.  K  From that perspective, I haven't thought much of the request.  Not saying uR others with different perspectives shouldn't have the tools they perceive needing.   Dave   -- h4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Roadl Vanderbilt, PA  15486h   ------------------------------  # Date: Thu, 23 Sep 2004 20:14:40 GMTe3 From: hammond@not@peek.ssr.hp.com (Charlie Hammond)RE Subject: Re: [DCL REQUEST] New ignore keyword for DIFFERENCES commandt2 Message-ID: <Q8G4d.11580$qt7.213@news.cpqcorp.net>   SOMEbody wrote:   G >> Alan's original post suggested the problem lies with lines that are dK >> comments.  As I understand it he said that DIFFERENCES with the various pL >> flags to ignore comments treated a line "$! This is a comment" as if the K >> line actually said "$" and could get itself confused and start matching ,K >> these lines with similarly treated comments at a different point in the   >> other file.    Ah, yes.  Been there. Done that.  F The workaround for this is to use /MATCH=x, where x is large enough to  prevent this sort of miss-match.  K The drawback to this is that you may reduce the numer if different sectionsd: and get sections with a lot of non-different lines listed.  I But I also agree with another comment: Comments are important.  Sometimes E the are as important or even more inportant than the code.  Why woulda6 you NOT want to examine the differences in comments???   -- eJ       Charlie Hammond -- Hewlett-Packard Company -- Ft Lauderdale  FL  USAF           (hammond@not@peek.ssr.hp.com -- remove "@not" when replying)J       All opinions expressed are my own and not necessarily my employer's.   ------------------------------    Date: 23 Sep 2004 20:02:18 -0700. From: spamsink2001@yahoo.com (Alan E. Feldman)E Subject: Re: [DCL REQUEST] New ignore keyword for DIFFERENCES commanda= Message-ID: <b096a4ee.0409231902.77d198a6@posting.google.com>i  n hammond@not@peek.ssr.hp.com (Charlie Hammond) wrote in message news:<r5A4d.11515$hV6.8250@news.cpqcorp.net>...C > Although I've DIF'd a lot of DCL procedures, I am having trouble  ; > understanding the problem here.  Did I miss and example? o& > If so, perhaps you would re-post it.  B I sometimes want to ignore comments in DCL command procedures. TheD problem is that you cannot ignore lines that comprise pure comments.A There are usually leading dollar signs left over. With lines like   D     $! Nothing but comments on this line. Another important comment.  ! we get DIFFERENCES output like...h       $            | $     $            | $     $            | $     $            | '     $            | )     $            | P     C What is the utility of displays like this? It wastes real estate onpE the screen. Also, the presence of identical lines of $'s often causes-B DIFFERENCES to lose synchronization. If these leading dollar signsB were also ignored, then one can have very clean DIFFERENCES output9 without having to resort to large /MATCH values to regain2 synchronization.  7 > I have found that /IGNORE=SPACING occasionally helps.  > Have you tried that?  D Yes, but it's not of much help. And I'm often doing this on VMS V6.1C in which /ignore=exact is broken, which means that ignoring spacinge compresses all the lines.    ------------------------------    Date: 23 Sep 2004 20:08:21 -0700. From: spamsink2001@yahoo.com (Alan E. Feldman)E Subject: Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command = Message-ID: <b096a4ee.0409231908.1ae4ac43@posting.google.com>e  j "Doc." <doc.cypher@openvms-rocks.com> wrote in message news:<Xns956DA0EE6E8C6dcovmsrox@212.100.160.123>...D > %NEWS-I-NEWMSG, Charlie Hammond wrote in news:r5A4d.11515$hV6.8250 > @news.cpqcorp.net  >  > > E > > Although I've DIF'd a lot of DCL procedures, I am having trouble  = > > understanding the problem here.  Did I miss and example?  ( > > If so, perhaps you would re-post it. > > 9 > > I have found that /IGNORE=SPACING occasionally helps.  > > Have you tried that? > F > Alan's original post suggested the problem lies with lines that are J > comments.  As I understand it he said that DIFFERENCES with the various K > flags to ignore comments treated a line "$! This is a comment" as if the aJ > line actually said "$" and could get itself confused and start matching J > these lines with similarly treated comments at a different point in the 
 > other file.e > . > I believe the behaviour he's seeking is thatH > 1)  Lines with ! comment delimiters are stripped back to "$". (normal)1 > 2)  Lines have the starting "$" stripped. (new)oJ > 3)  Comment lines that started "$!" are now blank and treated as thus by >     DIFFERENCES. >  >  > Doc.   By Jove you've got it! 0. @ 1. Strip the comment.R 2. Strip any trailing spaces.e 3. Strip the lone dollar sign." 4. Strip the remaining blank line!  6 So here's how it would look (with <sp> being a space):  ! 0. $<sp><sp>!  important commentst 1. $<sp><sp> 2. $ 3. <blank line>  4. <ignored blank line!>  B This way you are truly removing the comments. This way it works as? well as ignoring comments for other languages: the comments areo8 removed completely and only the actual code is compared.   ------------------------------    Date: 23 Sep 2004 20:10:07 -0700. From: spamsink2001@yahoo.com (Alan E. Feldman)E Subject: Re: [DCL REQUEST] New ignore keyword for DIFFERENCES commandI= Message-ID: <b096a4ee.0409231910.5a0b097b@posting.google.com>t  \ David Froble <davef@tsoft-inc.com> wrote in message news:<41532297.1030308@tsoft-inc.com>...
 > Doc. wrote:C > F > > %NEWS-I-NEWMSG, Charlie Hammond wrote in news:r5A4d.11515$hV6.8250 > > @news.cpqcorp.netg > >  > > E > >>Although I've DIF'd a lot of DCL procedures, I am having trouble d= > >>understanding the problem here.  Did I miss and example? n( > >>If so, perhaps you would re-post it. > >>9 > >>I have found that /IGNORE=SPACING occasionally helps.  > >>Have you tried that? > >> > > H > > Alan's original post suggested the problem lies with lines that are L > > comments.  As I understand it he said that DIFFERENCES with the various M > > flags to ignore comments treated a line "$! This is a comment" as if the nL > > line actually said "$" and could get itself confused and start matching L > > these lines with similarly treated comments at a different point in the  > > other file.n > > 0 > > I believe the behaviour he's seeking is thatJ > > 1)  Lines with ! comment delimiters are stripped back to "$". (normal)3 > > 2)  Lines have the starting "$" stripped. (new) L > > 3)  Comment lines that started "$!" are now blank and treated as thus by > >     DIFFERENCES. > >  > >  > > Doc. > >  > R > I know it's not part of the question/suggestion, but some of us, maybe just me, S > think comment lines are important.  If they are different, it might be important oM > to see the differences.  I have this quaint notion that any code should be -> > managable.  Out of date or errorneous comments can bite you.  B Yes, comments are important. If I thought they weren't, I wouldn't have them in my code to ignore!r  B Sometimes I do compare with comments included. Sometimes I want to? ignore them. Now isn't ignoring comments the whole point of the  /COMMENTS qualifier?   > M >  From that perspective, I haven't thought much of the request.  Not saying  T > others with different perspectives shouldn't have the tools they perceive needing. >  > Dave   ------------------------------    Date: 23 Sep 2004 20:21:20 -0700. From: spamsink2001@yahoo.com (Alan E. Feldman)E Subject: Re: [DCL REQUEST] New ignore keyword for DIFFERENCES command = Message-ID: <b096a4ee.0409231921.7e395b08@posting.google.com>   m hammond@not@peek.ssr.hp.com (Charlie Hammond) wrote in message news:<Q8G4d.11580$qt7.213@news.cpqcorp.net>...i > SOMEbody wrote:  > I > >> Alan's original post suggested the problem lies with lines that are lM > >> comments.  As I understand it he said that DIFFERENCES with the various EN > >> flags to ignore comments treated a line "$! This is a comment" as if the M > >> line actually said "$" and could get itself confused and start matching  M > >> these lines with similarly treated comments at a different point in the D > >> other file. > " > Ah, yes.  Been there. Done that. > H > The workaround for this is to use /MATCH=x, where x is large enough to" > prevent this sort of miss-match. > M > The drawback to this is that you may reduce the numer if different sections>< > and get sections with a lot of non-different lines listed.  E And you also end up with blocks containing nothing but leading dollard@ signs and | characters which are of no value and therefore waste screen real estate.>  K > But I also agree with another comment: Comments are important.  SometimesgG > the are as important or even more inportant than the code.  Why wouldg8 > you NOT want to examine the differences in comments???  ? Because sometimes I make changes in the code that don't requireaE changes in the comments. Like fixing a typo. Do I really want to listoF all my typo fixes in the comments? Or just simple debugging in which IE may have made formatting changes to the comments for various reasons.eE And such changes often wreak havoc with /IGNORE=(COMM,TRAI,BLAN). AndlF sometimes I have paragraphs of just general motivation that I need notF see every time I run DIFFERENCES after making spelling corrections and. format changes to said paragraphs of comments.  D And if there is no reason to ignore comments, then just why is there+ the /COMMENTS qualifier in the first place?o  D Hmmm, I don't want to edit out the comments with DIFFERENCES; I justD want to be able to ignore them properly in DCL files like I can with) files containing certain other languages!-   ------------------------------    Date: 23 Sep 2004 22:38:57 -0700. From: spamsink2001@yahoo.com (Alan E. Feldman)E Subject: Re: [DCL REQUEST] New ignore keyword for DIFFERENCES commandc= Message-ID: <b096a4ee.0409232138.4b157973@posting.google.com>a  m hammond@not@peek.ssr.hp.com (Charlie Hammond) wrote in message news:<Q8G4d.11580$qt7.213@news.cpqcorp.net>...e > SOMEbody wrote:f > I > >> Alan's original post suggested the problem lies with lines that are  M > >> comments.  As I understand it he said that DIFFERENCES with the various oN > >> flags to ignore comments treated a line "$! This is a comment" as if the M > >> line actually said "$" and could get itself confused and start matching  M > >> these lines with similarly treated comments at a different point in the o > >> other file. [...]a > K > But I also agree with another comment: Comments are important.  Sometimes G > the are as important or even more inportant than the code.  Why would 8 > you NOT want to examine the differences in comments???  C Another case of comments I want to ignore is commented out lines of.C code. These are not comments but are actual code lines, though theyoE are inactive. And sometimes I want to see the differences without thepF extra clutter of these commented out lines. Sometimes I do want to see them. But if I use justsD /IGNORE=(COMMENTS,TRAILING_SPACES,BLANK_LINES), I get the problem of> blocks of dollar signs that waste screen space and the need to increase the match value.    ------------------------------   End of INFO-VAX 2004.530 ************************