1 INFO-VAX	Thu, 29 Jun 2006	Volume 2006 : Issue 359       Contents:* Re: CSWS 2.1 Startup failure after upgrade> Re: Education Ministry rethinks payroll plans (Oracle and VMS) Re: Floating point questions Re: Floating point questions Re: Floating point questions Re: Floating point questions Re: Floating point questions Re: Floating point questions Re: Floating point questions Itanic in the news (june-29)F Re: ntpdate fails after TCPIP V5.5 upgrade RESOLVED (was: 8.2 upgrade)' Re: OT: Intel quad core X64 benchmarked ' Re: OT: Intel quad core X64 benchmarked ' Re: OT: Intel quad core X64 benchmarked 6 Re: Python 2.5b1 is available for OpenVMS AXP and IA64 Re: SFTP Output redirection  Re: SFTP Output redirection  Re: SFTP Output redirection  Re: SFTP Output redirection 
 VMS Backup Re: VMS Backup Re: VMS Backup Re: VMS Backup Re: VMS Backup Re: VMS Backup Re: VMS Backup Re: VMS BackupG Re: zipping large files (was Re: Info-ZIP's Zip V2.32 is now available) G Re: zipping large files (was Re: Info-ZIP's Zip V2.32 is now available)   F ----------------------------------------------------------------------  # Date: Thu, 29 Jun 2006 17:19:36 GMT , From: "warren sander" <warren.sander@hp.com>3 Subject: Re: CSWS 2.1 Startup failure after upgrade - Message-ID: <IYTog.10$rI4.8@news.cpqcorp.net>   K it would be good to know what is on line 155 of httpd.conf. That might give  a clue. H looks like something with ip address, virtualhost, listen directive. but line 155 might give a clue  $ <Uwe.Kroyer@gmx.de> wrote in message< news:1151253218.900363.85870@c74g2000cwc.googlegroups.com...H > I've upgraded CSWS 2.0 to CSWS 2.1 on a XP1000 with OpenVMS 7.3-2, allI > available patches installed. TCPIP Version is 5.4-15. After the upgrade  > CSWS startup fails with: > G >        [SUN JUN25 17:48:17 2006] [crit] (6)no such device or address:  >        alloc_listener:+ >        failed to get a socket for 0.0.0.0 I >        syntax error on line 155 of /apache$root/000000/conf/httpd.conf:  >        Listen setup failed > D > CSWS 2.0 has showed the same failure but during CSWS restart only. > ( > I start CSWS under the SYSTEM account. >  > Any help is appreciated  >    ------------------------------    Date: 29 Jun 2006 04:25:01 -0700 From: dooleys@snowy.net.auG Subject: Re: Education Ministry rethinks payroll plans (Oracle and VMS) C Message-ID: <1151580301.640198.307760@m73g2000cwd.googlegroups.com>   H > The bigger question might be "Can the Customers and Vendors afford theJ > resources and effort associated with the QA testing of their applicationI > with the 10-20 monthly *security* (not bug) OS and supporting utilities 2 > patches associated with Linux and Windows OS's?" > A > [RH Security site lists approx 7-20 security patches per month] B The answer is that software suppliers can always do QA and testingD (often of variable quality) if they have a significant market share.D I have little experience of applications on linux, but most software	 suppliers C will run regression tests for major versions of windows and service  packs.G As regards security patches, the software suppliers get windows updates E at the same time as their customers, so the customer just has to hope  that@ nothing breaks, and if it does, uninstall the most recent patch!E In practice, there is significantly more benefit than pain in keeping 
 up to dateF with security patches, my experience is that most application software will just continue to work.F (I have also seen this occur in vms environments, application software@ would be regression tested for vms version upgrades, or compiler@ upgrades, but a mandatory patch could be applied to a production: system - as long as a system disk backup was taken first!)B > In addition, can the Cust afford to deal with HR security issuesG > associated with hackers getting access to Cust and employee personnel  > data?  > E > For some vendors and Cust's implementing IT systems, this is a huge G > consideration - especially since the focus going forward is web based $ > services for end users or clients.C Yes this is a major risk with an organisation's HR/Payroll, but the C methods of securing these systems is much the same, irrespective of  the operating system in use.G > As an example, how much longer would you deal with a bank or firm who I > just sent you a letter stating that under the State law, they needed to F > inform you that all the personal financial, home address and contact5 > information they had on you is now on the Internet? ) Personally it would not bother me at all! J > While no platform is 100% secure, imho, it stands to reason that puttingJ > your HR system on a platform that has 10-20 security patches *per month*G > (e.g. Linux - Windows is similar) is obviously placing your firm at a H > higher risk than a platform that has much, much fewer security issues.5 and that is why vms runs x% of HR systems  worldwide.  and what is x? 2? 3?0 vms may be an excellent choice for some systems," but HR/payroll is not one of them. Phil   ------------------------------   Date: 29 Jun 2006 11:32:32 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)% Subject: Re: Floating point questions , Message-ID: <4ghs2gF1n9ichU1@individual.net>  9 In article <pbmdnWPxweiwrD7ZnZ2dnUVZ_rSdnZ2d@libcom.com>, * 	Dave Froble <davef@tsoft-inc.com> writes: > JF Mezei wrote: " >> briggs@encompasserve.org wrote:J >>> Doesn't help if you have to convert from float binary to float decimal >>> to perform the comparison. >>> ; >>> We haven't been told what format the source data is in.  >>  F >> Source data is a GPS that provides litten endian IEEE 32 bit value. >>  K >> Someone mentioned that 1.0e25 could not be precisely represented in such G >> a 32 bit entity ? Is that really the case ?  Isn't it just a case of E >> representing 1.0 and then "25" in the field that defines where the  >> decimal is located ?  > D > In your particular case you're dealing in latitude and longitude, H > usually expressed in degrees, minutes, seconds, and fractions thereof. > J > It will not be hard to convert such to whatever measurement you wish to ) > use, meters, feet, yards, miles, etc.      Not hard????   What is the distance between: -     70d0m0s E 70d0m0s N and 80d0m0s E 70d0m0s  and now:-     70d0m0s E 10d0m0s N and 80d0m0s E 10d0m0s   = Somehow I have to believe the math needed to compute distance @ based on latitude and longitude is complex enough to not qualify as "not hard".  :-) A Of course, if there is a simple method I would love to see it.  I ? assumed the this difficulty is why the military always used UTM  rather than Lat/Lon.   bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------    Date: 29 Jun 2006 06:37:21 -0500 From: briggs@encompasserve.org% Subject: Re: Floating point questions 3 Message-ID: <Car1RQ0nY2Rs@eisner.encompasserve.org>   \ In article <44A329D7.9D9DA363@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes:! > briggs@encompasserve.org wrote: I >> Doesn't help if you have to convert from float binary to float decimal  >> to perform the comparison.  >>  : >> We haven't been told what format the source data is in. > E > Source data is a GPS that provides litten endian IEEE 32 bit value.  > J > Someone mentioned that 1.0e25 could not be precisely represented in suchF > a 32 bit entity ? Is that really the case ?  Isn't it just a case ofD > representing 1.0 and then "25" in the field that defines where the > decimal is located ?  B 1.0E25 cannot be represented in a 32 binary floating point format.D IEEE floating point is an example of a binary floating point format.A So it is REALLY the case that 1.0E25 is not exactly representable  in 32 bit IEEE floating point.  ? All the usual binary floating representation schemes are slight " variations on the following theme:   You have n bits of exponent " You have m bits of binary fraction You have a sign bit   - You can encode exactly any number of the form   6 +/- 0.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx * 2^yyyyyyyyyyyyy=       ( m bits of fraction               ( n bits of exponent @         expressed in binary )              expressed in binary )  E [Variations include support for normalized or denormalized fractions, @ inclusion of the redundant leading bit in the case of normalizedF fractions, hexadecimal exponents, the precise range of exponent values> and their encoding, field placement in the encoded bit string,5 special codes for "infinity" and "not a number", etc]   1 If you are doing 32 bit floating point, m+n <= 32   	 1.0E25 is   D 0.10000100010110010101000101100001010000000001010010000100101 * 2^84  ?   (barring any errors on my part -- but it's within a bit or 2)    And   ; 10000100010110010101000101100001010000000001010010000100101   < won't fit into a binary field that is less than 59 bits long  - You can't do that with 32 bit floating point.   D You can't do that with 64 bit floating point and still have a decent range of exponents.   . You _can_ do that with 128 bit floating point.  K As has been pointed out here you also _can_ do that with a decimal floating : point scheme.  But IEEE floating point is a binary scheme.   ------------------------------  % Date: Thu, 29 Jun 2006 09:31:50 -0400 ' From: Dave Froble <davef@tsoft-inc.com> % Subject: Re: Floating point questions 9 Message-ID: <ZrydnXatrricSD7ZnZ2dnUVZ_q-dnZ2d@libcom.com>    Bill Gunshannon wrote:; > In article <pbmdnWPxweiwrD7ZnZ2dnUVZ_rSdnZ2d@libcom.com>, , > 	Dave Froble <davef@tsoft-inc.com> writes: >> JF Mezei wrote:# >>> briggs@encompasserve.org wrote: K >>>> Doesn't help if you have to convert from float binary to float decimal  >>>> to perform the comparison.  >>>>< >>>> We haven't been told what format the source data is in.G >>> Source data is a GPS that provides litten endian IEEE 32 bit value.  >>> L >>> Someone mentioned that 1.0e25 could not be precisely represented in suchH >>> a 32 bit entity ? Is that really the case ?  Isn't it just a case ofF >>> representing 1.0 and then "25" in the field that defines where the >>> decimal is located ?E >> In your particular case you're dealing in latitude and longitude,  I >> usually expressed in degrees, minutes, seconds, and fractions thereof.  >>K >> It will not be hard to convert such to whatever measurement you wish to  * >> use, meters, feet, yards, miles, etc.   >  > Not hard???? >  > What is the distance between: / >     70d0m0s E 70d0m0s N and 80d0m0s E 70d0m0s 
 > and now:/ >     70d0m0s E 10d0m0s N and 80d0m0s E 10d0m0s  > ? > Somehow I have to believe the math needed to compute distance B > based on latitude and longitude is complex enough to not qualify > as "not hard".  :-) C > Of course, if there is a simple method I would love to see it.  I A > assumed the this difficulty is why the military always used UTM  > rather than Lat/Lon. >  > bill >   A Assume a sphere.  Ok, I'm aware that the Planet is not really so  : uniform, but for most purposes the assumption is adequate.  I There was a time when I'd have been able to do the math.  That was about  G 40 years ago, and I'm not going to dig out the books for this question.   G I've got a couple under $100 GPS units that will give me distance from  $ one point to another.  It is doable.  E Now if altitude is also a consideration, then the math is a bit more  G complex.  However, once the formulas are set up, it's just a matter of   plugging in the numbers.   --  4 David Froble                       Tel: 724-529-0450> Dave Froble Enterprises, Inc.      E-Mail: davef@tsoft-inc.com DFE Ultralights, Inc.  170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Thu, 29 Jun 2006 07:08:06 -0700 # From: "Tom Linden" <tom@kednos.com> % Subject: Re: Floating point questions ) Message-ID: <op.tbwuvspwzgicya@hyrrokkin>   I On Wed, 28 Jun 2006 21:17:36 -0700, Steven M. Schweda <sms@antinode.org>=    =    wrote:  A >    So, does this count as the annual comp.os.vms floating-point   > mystification thread for 2006?  B A good candidate.  Decimal and binary don't mix well.  If you need@ to use Decimal, then use a language that supports it as a native
 data type.   ------------------------------  # Date: Thu, 29 Jun 2006 15:57:34 GMT % From: Rob Brown <mylastname@gmcl.com> % Subject: Re: Floating point questions E Message-ID: <Pine.LNX.4.61.0606290951250.15570@localhost.localdomain>   * On Thu, 29 Jun 2006, Karsten Nyblad wrote:   > JF Mezei wrote: / >>> So you could theoretically have 8388607e255  >>   >>  C >> OK, read a bit more. 255 as exponent is a special value. (255 as G >> exponent and 0 as fraction means "infinity".  So 254 is the highest  1 >> exponent to be used. Still pretty high number.  >>  > >> http://www.psc.edu/general/software/packages/ieee/ieee.html > C > No you seem to think that the exponent denotes the potents of 10  B > that the mantissa should be multiplied with.  It does not.  The @ > exponent denotes the potents of 2 that the mantissa should be  > multiplied with. > B > E.g.  You can get a precise representation of 31457 and 31457 * $ > 2**200, but not of 31457 * 10**50.  G In fact, you can't even represent 31457 * 2**200, because the range of   the exponent is -126 to +127.      --    B Rob Brown                        b r o w n a t g m c l d o t c o m6 G. Michaels Consulting Ltd.      (780)438-9343 (voice)4 Edmonton                         (780)437-3367 (FAX)2                                   http://gmcl.com/   ------------------------------    Date: 29 Jun 2006 09:14:11 -0700- From: "Doug Phillips" <dphill46@netscape.net> % Subject: Re: Floating point questions C Message-ID: <1151597650.855546.190310@y41g2000cwy.googlegroups.com>    Bill Gunshannon wrote:; > In article <ZrydnXatrricSD7ZnZ2dnUVZ_q-dnZ2d@libcom.com>, , > 	Dave Froble <davef@tsoft-inc.com> writes: > > Bill Gunshannon wrote:> > >> In article <pbmdnWPxweiwrD7ZnZ2dnUVZ_rSdnZ2d@libcom.com>,/ > >> 	Dave Froble <davef@tsoft-inc.com> writes:  > >>> JF Mezei wrote: & > >>>> briggs@encompasserve.org wrote:L > >>>>> Doesn't help if you have to convert from float binary to float deci= mal " > >>>>> to perform the comparison. > >>>>> ? > >>>>> We haven't been told what format the source data is in. J > >>>> Source data is a GPS that provides litten endian IEEE 32 bit value. > >>>>L > >>>> Someone mentioned that 1.0e25 could not be precisely represented in = suchK > >>>> a 32 bit entity ? Is that really the case ?  Isn't it just a case of I > >>>> representing 1.0 and then "25" in the field that defines where the  > >>>> decimal is located ? G > >>> In your particular case you're dealing in latitude and longitude, L > >>> usually expressed in degrees, minutes, seconds, and fractions thereof. > >>> L > >>> It will not be hard to convert such to whatever measurement you wish = to+ > >>> use, meters, feet, yards, miles, etc.  > >> > >> Not hard????  > >>" > >> What is the distance between:2 > >>     70d0m0s E 70d0m0s N and 80d0m0s E 70d0m0s
 > >> and now: 2 > >>     70d0m0s E 10d0m0s N and 80d0m0s E 10d0m0s > >>B > >> Somehow I have to believe the math needed to compute distanceE > >> based on latitude and longitude is complex enough to not qualify  > >> as "not hard".  :-)F > >> Of course, if there is a simple method I would love to see it.  ID > >> assumed the this difficulty is why the military always used UTM > >> rather than Lat/Lon.  > >>	 > >> bill  > >> > > D > > Assume a sphere.  Ok, I'm aware that the Planet is not really so> > > uniform, but for most purposes the assumption is adequate. > > L > > There was a time when I'd have been able to do the math.  That was aboutK > > 40 years ago, and I'm not going to dig out the books for this question.  > > J > > I've got a couple under $100 GPS units that will give me distance from( > > one point to another.  It is doable. > @ > I wasn't arguing that it wasn't doable.  Just that because theB > distance between lines of longitude are not uniform varying with@ > latitude the math is more complex.  When someone tells me someC > math problem is "not hard" I assume that means the average person D > can do it nn their head.  Somehow I doubt that the linear distanceB > between two arbitrary points on the globe meets that definition. >  > > H > > Now if altitude is also a consideration, then the math is a bit moreJ > > complex.  However, once the formulas are set up, it's just a matter of > > plugging in the numbers. > D > But that's true about any math formula no matter how complex.  TheB > initial programming is the part that doesn't meet the "not hard" > criteria.  >     Look up "great circle distances"  5 Assume Earth is a perfect sphere of radius 6371.2 km: ? Convert longitude and latitude to radians (multiply by pi/180),  then compute as follows:    theta =3D lon2 - lon1H  distance =3D acos(sin(lat1) =D7 sin(lat2) + cos(lat1) =D7 cos(lat2) =D7 cos(theta)) /  if  (distance < 0)  distance =3D distance + pi !  distance =3D distance =D7 6371.2   A The "perfect sphere" assumption yields very close approximations. 8 Variant tables would be used to obtain greater accuracy.  @ In math programming terms, I consider "not hard" to mean there'sF existing code or an existing formula you can convert so you don't needD to invent one. I was tasked to write a great circle distance programE once on a computer that had only 13 assembler instructions (no higher C language) with only integer decimal math. Doing asin,sin,cos *were*  challenges.    ------------------------------    Date: 29 Jun 2006 10:22:25 -0700- From: "Doug Phillips" <dphill46@netscape.net> % Subject: Re: Floating point questions C Message-ID: <1151601745.879651.232570@b68g2000cwa.googlegroups.com>    Doug Phillips wrote: > Bill Gunshannon wrote:= > > In article <ZrydnXatrricSD7ZnZ2dnUVZ_q-dnZ2d@libcom.com>, . > > 	Dave Froble <davef@tsoft-inc.com> writes: > > > Bill Gunshannon wrote:@ > > >> In article <pbmdnWPxweiwrD7ZnZ2dnUVZ_rSdnZ2d@libcom.com>,1 > > >> 	Dave Froble <davef@tsoft-inc.com> writes:  > > >>> JF Mezei wrote: ( > > >>>> briggs@encompasserve.org wrote:L > > >>>>> Doesn't help if you have to convert from float binary to float de= cimal $ > > >>>>> to perform the comparison.	 > > >>>>> A > > >>>>> We haven't been told what format the source data is in. L > > >>>> Source data is a GPS that provides litten endian IEEE 32 bit value. > > >>>>L > > >>>> Someone mentioned that 1.0e25 could not be precisely represented i= n suchL > > >>>> a 32 bit entity ? Is that really the case ?  Isn't it just a case = ofK > > >>>> representing 1.0 and then "25" in the field that defines where the  > > >>>> decimal is located ? I > > >>> In your particular case you're dealing in latitude and longitude, L > > >>> usually expressed in degrees, minutes, seconds, and fractions there= of.  > > >>> L > > >>> It will not be hard to convert such to whatever measurement you wis= h to- > > >>> use, meters, feet, yards, miles, etc.  > > >> > > >> Not hard????  > > >>$ > > >> What is the distance between:4 > > >>     70d0m0s E 70d0m0s N and 80d0m0s E 70d0m0s > > >> and now: 4 > > >>     70d0m0s E 10d0m0s N and 80d0m0s E 10d0m0s > > >>D > > >> Somehow I have to believe the math needed to compute distanceG > > >> based on latitude and longitude is complex enough to not qualify  > > >> as "not hard".  :-)H > > >> Of course, if there is a simple method I would love to see it.  IF > > >> assumed the this difficulty is why the military always used UTM > > >> rather than Lat/Lon.  > > >> > > >> bill  > > >> > > > F > > > Assume a sphere.  Ok, I'm aware that the Planet is not really so@ > > > uniform, but for most purposes the assumption is adequate. > > > L > > > There was a time when I'd have been able to do the math.  That was ab= out L > > > 40 years ago, and I'm not going to dig out the books for this questio= n=2E > > > L > > > I've got a couple under $100 GPS units that will give me distance from* > > > one point to another.  It is doable. > > B > > I wasn't arguing that it wasn't doable.  Just that because theD > > distance between lines of longitude are not uniform varying withB > > latitude the math is more complex.  When someone tells me someE > > math problem is "not hard" I assume that means the average person F > > can do it nn their head.  Somehow I doubt that the linear distanceD > > between two arbitrary points on the globe meets that definition. > >  > > > J > > > Now if altitude is also a consideration, then the math is a bit moreL > > > complex.  However, once the formulas are set up, it's just a matter of > > > plugging in the numbers. > > F > > But that's true about any math formula no matter how complex.  TheD > > initial programming is the part that doesn't meet the "not hard"
 > > criteria.  > >  > " > Look up "great circle distances" > 7 > Assume Earth is a perfect sphere of radius 6371.2 km: A > Convert longitude and latitude to radians (multiply by pi/180),  > then compute as follows: >  >  theta =3D lon2 - lon1J >  distance =3D acos(sin(lat1) =D7 sin(lat2) + cos(lat1) =D7 cos(lat2) =D7
 > cos(theta)) 1 >  if  (distance < 0)  distance =3D distance + pi # >  distance =3D distance =D7 6371.2  > C > The "perfect sphere" assumption yields very close approximations. : > Variant tables would be used to obtain greater accuracy. > B > In math programming terms, I consider "not hard" to mean there'sH > existing code or an existing formula you can convert so you don't needF > to invent one. I was tasked to write a great circle distance programG > once on a computer that had only 13 assembler instructions (no higher E > language) with only integer decimal math. Doing asin,sin,cos *were* 
 > challenges.     F Forgot the disclaimer: The example formula presented in my quoted postE is not a complete presentation of all calculations required for great @ circle navigation and should *not* be used for actual navigation	 purposes.   D (in case someone actually thought it was and decided to use it as is> without understanding more about the earth and the formula :-)   ------------------------------  % Date: Thu, 29 Jun 2006 06:46:52 -0400 ) From: "Neil Rieck" <n.rieck@sympatico.ca> % Subject: Itanic in the news (june-29) < Message-ID: <44a3af1d$0$18509$9a6e19ea@news.newshosting.com>   Since this is fact:   ?    Marvell slumps after $600M purchase in Intel's comm division 8    http://www.theinquirer.net/default.aspx?article=32692  % Then there may be some truth to this:     Itanic in the news (june-29) 8    http://www.theinquirer.net/default.aspx?article=32714  
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada." http://www3.sympatico.ca/n.rieck/    ------------------------------    Date: 29 Jun 2006 00:04:31 -0700- From: "Wilm Boerhout" <w5.boerhout@planet.nl> O Subject: Re: ntpdate fails after TCPIP V5.5 upgrade RESOLVED (was: 8.2 upgrade) C Message-ID: <1151564671.553842.207250@x69g2000cwx.googlegroups.com>    Volker Halle wrote:   G > I've seen this error on TCPIP V5.5, when the NTP service had not been E > configured at least once (does not need to be started). On previous @ > version, NTPDATE seems to work, even if the NTP serivce is not
 > configured.   C This hit the spot, Volker. I configured and enabled the ntp service D (but did not start it). After that, the ntpdate command worked OK. I= then disabled the ntp service, and ntpdate continued to work.   E So what remains is an omission in the TCPIP V5.5 release notes, like: D "In order for the ntpdate command to work, you need to configure and then disable the ntp service."   Thanks!    /Wilm    ------------------------------    Date: 29 Jun 2006 05:28:04 -0700- From: "Andrew" <andrew_harrison@symantec.com> 0 Subject: Re: OT: Intel quad core X64 benchmarkedA Message-ID: <1151584084.213485.50370@75g2000cwc.googlegroups.com>    JF Mezei wrote:  > Andrew wrote: H > > 3.0 Ghz Xeon 5160 does 3057 SPEcint2005 vs 1.6 Ghz 9MB IA-64 at 15900 > > SPECint and 2797 SPECfp vs 2712 for Itanium. > > I > > It would appear that with the introduction of Woodcrest that the last - > > refuge for Itanium, FP has been breached.  >  > F > In fairness, that IA64 thing is expected to have its next generationH > soon, so performance of the new 8086s should be compared once they get( > that new IA64 thing out on the market. > F > Also, for large systems, raw CPU performance is only one part of theG > equation. The ability to exchange data with the rest of the system is  > equally important. > I > Does IA64 still hold any advantage in terms of memory/bus interconnects J > and scalability ? Or has the 8086 truly overtaken IA64 on all fronts now > ?   @ It depends, Xeon 5160 and Itanium II both use the Frontside bus,G Itanium II supports up to 667 Mhz Frontside while Xeon 5160 supports up 7 to 1333 Mhz roughly double the bandwidth of Itanium II.   C So for small 1-4 Module systems which just use Frontside Xeon has a E clear lead and this is probably demonstrated by the Xeon's SPECratefp B result where the Itaniums small single core lead evaporates when 4 cores are used.   G Of course you have to use something else appart from Frontside when you C go beyond 4 modules and Itanium implimenentations probably have the D edge over Xeon at this point, however there is no reason inherant to! Xeon why this has to be the case.    Regards  Andrew Harrison    ------------------------------    Date: 29 Jun 2006 08:47:38 -0700 From: icerq4a@spray.se0 Subject: Re: OT: Intel quad core X64 benchmarkedB Message-ID: <1151596058.896458.94410@p79g2000cwp.googlegroups.com>  
 Andrew skrev:    > JF Mezei wrote:  > > Andrew wrote: J > > > 3.0 Ghz Xeon 5160 does 3057 SPEcint2005 vs 1.6 Ghz 9MB IA-64 at 15902 > > > SPECint and 2797 SPECfp vs 2712 for Itanium. > > > K > > > It would appear that with the introduction of Woodcrest that the last / > > > refuge for Itanium, FP has been breached.  > >  > > H > > In fairness, that IA64 thing is expected to have its next generationJ > > soon, so performance of the new 8086s should be compared once they get* > > that new IA64 thing out on the market. > > H > > Also, for large systems, raw CPU performance is only one part of theI > > equation. The ability to exchange data with the rest of the system is  > > equally important. > > K > > Does IA64 still hold any advantage in terms of memory/bus interconnects L > > and scalability ? Or has the 8086 truly overtaken IA64 on all fronts now > > ?  > B > It depends, Xeon 5160 and Itanium II both use the Frontside bus,I > Itanium II supports up to 667 Mhz Frontside while Xeon 5160 supports up 9 > to 1333 Mhz roughly double the bandwidth of Itanium II.   A As usual, one should not expect accurate information from Andrew.   E Itanium 2 use a double pumped 128-bit wide FSB while Xeon uses a quad E pumped 64-bit wide FSB. A 667 Mhz Itanium FSB have the same bandwidth  as a 1333Mhz Xeon FSB.  E > So for small 1-4 Module systems which just use Frontside Xeon has a G > clear lead and this is probably demonstrated by the Xeon's SPECratefp D > result where the Itaniums small single core lead evaporates when 4 > cores are used.   C Xeon does not have a clear lead. HP's xz1 scale good to 4 cores and 6 SGIs system scales linear from 1 to hundreds of cores.   ------------------------------  % Date: Thu, 29 Jun 2006 09:08:45 -0700 # From: "Tom Linden" <tom@kednos.com> 0 Subject: Re: OT: Intel quad core X64 benchmarked) Message-ID: <op.tbw0gv0rzgicya@hyrrokkin>   = On Thu, 29 Jun 2006 08:47:38 -0700, <icerq4a@spray.se> wrote:    >  > Andrew skrev:  >  >> JF Mezei wrote: >> > Andrew wrote:I >> > > 3.0 Ghz Xeon 5160 does 3057 SPEcint2005 vs 1.6 Ghz 9MB IA-64 at 1=  590 3 >> > > SPECint and 2797 SPECfp vs 2712 for Itanium.  >> > >I >> > > It would appear that with the introduction of Woodcrest that the =   =   >> last 0 >> > > refuge for Itanium, FP has been breached. >> > >> >I >> > In fairness, that IA64 thing is expected to have its next generatio=  n I >> > soon, so performance of the new 8086s should be compared once they =  get + >> > that new IA64 thing out on the market.  >> >I >> > Also, for large systems, raw CPU performance is only one part of th=  e I >> > equation. The ability to exchange data with the rest of the system =  is >> > equally important.  >> >A >> > Does IA64 still hold any advantage in terms of memory/bus  =    >> interconnectsI >> > and scalability ? Or has the 8086 truly overtaken IA64 on all front=  s  =   >> now >> > ? >>C >> It depends, Xeon 5160 and Itanium II both use the Frontside bus, I >> Itanium II supports up to 667 Mhz Frontside while Xeon 5160 supports =  up: >> to 1333 Mhz roughly double the bandwidth of Itanium II. > C > As usual, one should not expect accurate information from Andrew.  > G > Itanium 2 use a double pumped 128-bit wide FSB while Xeon uses a quad G > pumped 64-bit wide FSB. A 667 Mhz Itanium FSB have the same bandwidth  > as a 1333Mhz Xeon FSB.  G I suspect the utilization of the FSB would be higher on XEON, since the . wider bus will likely contain extraneous data.   > F >> So for small 1-4 Module systems which just use Frontside Xeon has aI >> clear lead and this is probably demonstrated by the Xeon's SPECratefp=   E >> result where the Itaniums small single core lead evaporates when 4  >> cores are used. > E > Xeon does not have a clear lead. HP's xz1 scale good to 4 cores and 8 > SGIs system scales linear from 1 to hundreds of cores. >  SGI is dropping Itanium.   ------------------------------    Date: 29 Jun 2006 09:33:06 -0700( From: "Rich Jordan" <jordan@ccs4vms.com>? Subject: Re: Python 2.5b1 is available for OpenVMS AXP and IA64 B Message-ID: <1151598786.453647.32670@p79g2000cwp.googlegroups.com>  ! Jean-Fran=E7ois Pi=E9ronne wrote: I > A Python 2.5 kit is now available for OpenVMS AXP and IA64, this kit is  > based on Python 2.5b1. >  > L > It include some new libraries like, for example, pyExcelerator, Construct. > @ > Some new LIB$ and SYS$ routines interfaces have also be added. > J > More information on Python from the official site http://www.python.org/L > Download OpenVMS version from http://www.pi-net.dyndns.org/anonymous/kits/ > : > Various demonstrations from http://vmspython.dyndns.org/ >  >  > Jean-Fran=E7ois   G Thanks for all the work, Jean-Fran=E7ois; looking forward to trying it.    Rich CCS    ------------------------------    Date: 29 Jun 2006 01:01:54 -0700* From: "Palda" <jiri.pallich@logicacmg.com>$ Subject: Re: SFTP Output redirectionC Message-ID: <1151568114.362963.304360@b68g2000cwa.googlegroups.com>    Hi Wilm,  E thanks a lot for your response. I have tried what you have advised to F me, but the behaviour has not changed. I still get the warning messageE about cursor movement capability and the output is still written on a 	 screen :(    Best regards Jiri     Jir=ED Pallich SMSC Team / Developer ( Global Telecoms Products, Czech Republic) _________________________________________   	 LogicaCMG    Holandsk=E1 5, Brno  Czech Republic T: +420 543 125 368  E: jiri.pallich@logicacmg.com  www.logicacmg.com          Wilm Boerhout napsal:  > jiri,  > D > Don't know much about sftp, but from the looks of it, it seems theH > program is written with interactive use on a VT-like terminal in mind. > C > It could be that it assigns a channel to the terminal device "TT" ) > itself, so you could try the following:  > ( > $ DEFINE /USER SYS$OUTPUT SFTP_CMD.LOG > $ DEFINE /USER TT NL: & > $ SFTP "-B" SFTP_CMD.ARG "user@node" > E > (note that the /USER saves you from the $ DEASSIGN SYS$OUTPUT stuf, - > image rundown clears the logicals as well.)  >=20 > HTH, > Wilm   ------------------------------    Date: 29 Jun 2006 01:04:00 -0700* From: "Palda" <jiri.pallich@logicacmg.com>$ Subject: Re: SFTP Output redirectionC Message-ID: <1151568240.606253.313420@b68g2000cwa.googlegroups.com>    Hi Ian,   7 I am using TCPIP ver 5.4 - ECO 5, is it the latest one?    Regards  Jiri     Jir=ED Pallich SMSC Team / Developer ( Global Telecoms Products, Czech Republic) _________________________________________   	 LogicaCMG    Holandsk=E1 5, Brno  Czech Republic T: +420 543 125 368  E: jiri.pallich@logicacmg.com  www.logicacmg.com        Ian Miller napsal:E > Make sure you are on TCPIP V5.4 and latest ECO kit. There have been  > lots of fixes for ssh.   ------------------------------    Date: 29 Jun 2006 07:15:19 -0700* From: "Palda" <jiri.pallich@logicacmg.com>$ Subject: Re: SFTP Output redirectionC Message-ID: <1151590519.307722.184100@d56g2000cwd.googlegroups.com>    I'm afraid it does :(    Jiri   ------------------------------  % Date: Thu, 29 Jun 2006 19:13:09 +0200 / From: Paul Sture <paul.sture.nospam@hispeed.ch> $ Subject: Re: SFTP Output redirection: Message-ID: <44599$44a40a26$50db5015$1628@news.hispeed.ch>   Palda wrote: > I'm afraid it does :(  >  > Jiri >  Then it's a definite bug.   4 Do you have support channels you can report this to?   ------------------------------  % Date: Thu, 29 Jun 2006 10:34:03 -0400 + From: Chuck Aaron <caaron@ceris.purdue.edu>  Subject: VMS Backup 5 Message-ID: <e80ocs$ns7$1@mailhub227.itcs.purdue.edu>   I If I am sitting in a directory called user$disk:[httpd.aapco] and I want  * to backup everything in the .bin directoryG tree under aapco, what command would I use to insure the bin directory  ( and all directories and files underneath$ bin maintained their existing owner?  I For instance, let's say I wanted to use the backup command to save it to   a saveset called aapcobin.bck.         Also,   @ once I copy the *.bck file to the new node and it is sitting in , web$disk:[httpd.aapco] what command do I useG to restore it with replace capability to web$disk:[httpd.aapco.bin...]  9 and keep the original owner of the directories and files?    Thank you in advance.    ------------------------------    Date: 29 Jun 2006 07:49:53 -0700  From: "Ian Miller" <ijm@uk2.net> Subject: Re: VMS Backup B Message-ID: <1151592593.779796.148690@75g2000cwc.googlegroups.com>   I think   $ $ BACKUP [.bin...] aapcobin.bck/save  5 $ BACKUP aapcobin.bck/save [.bin...]/OWN=ORIG/REPLACE    ------------------------------    Date: 29 Jun 2006 08:17:17 -0700$ From: "AEF" <spamsink2001@yahoo.com> Subject: Re: VMS Backup B Message-ID: <1151594237.760778.313930@75g2000cwc.googlegroups.com>   Chuck Aaron wrote:J > If I am sitting in a directory called user$disk:[httpd.aapco] and I want, > to backup everything in the .bin directoryH > tree under aapco, what command would I use to insure the bin directory* > and all directories and files underneath& > bin maintained their existing owner? > J > For instance, let's say I wanted to use the backup command to save it to  > a saveset called aapcobin.bck.    = $ BACKUP USER$DISK:[HTTPD.AAPCO.BIN...] AAPCOBIN.BCK/SAVE_SET   C plus any other qualifiers you normally use. No special qualifier is . needed to save the owner -- it's already done.     > Also,  > A > once I copy the *.bck file to the new node and it is sitting in . > web$disk:[httpd.aapco] what command do I useH > to restore it with replace capability to web$disk:[httpd.aapco.bin...]; > and keep the original owner of the directories and files?     6 $ BACKUP AAPCOBIN.BCK [*...]/BY_OWNER=ORIGINAL/REPLACE    F With VMS BACKUP, [*...] is "secret code" that means "restore the filesA to the same directories they were saved from, but on the "current F device"". Be advised that this will restore *all* files present in theF save set. If the save set had files from [HTTPD.AAPCO...], e.g., you'd< have to add /SELECT=[HTTPD.AAPCO.BIN...] after the save-set.  E You *do* need a qualifier to restore to the original owner except for @ /image, /incremental, and /physical restore operations. And that% qualifier is the /BY_OWNER qualifier.   G Please read the documentation for BACKUP in the System Utilities manual  for more information.      >  > Thank you in advance.    AEF    ------------------------------  % Date: Thu, 29 Jun 2006 11:37:31 -0400 + From: Chuck Aaron <caaron@ceris.purdue.edu>  Subject: Re: VMS Backup 5 Message-ID: <e80s3r$pn0$1@mailhub227.itcs.purdue.edu>   D SYSTEM> BACKUP/confirm AAPCOBIN.BCK [*...]/BY_OWNER=ORIGINAL/REPLACE5 WEB$DISK:[HTTPD.AAPCO]AAPCOBIN.BCK;1, copy? (Y or N):   F with the above statement, it is only going to put the first file into ! web$disk:[httpd.aapco] instead of , web$disk:[httpd.aapco.bin] where it belongs.   Chuck   
 AEF wrote:   >Chuck Aaron wrote:  >    > J >>If I am sitting in a directory called user$disk:[httpd.aapco] and I want, >>to backup everything in the .bin directoryH >>tree under aapco, what command would I use to insure the bin directory* >>and all directories and files underneath& >>bin maintained their existing owner? >>J >>For instance, let's say I wanted to use the backup command to save it to  >>a saveset called aapcobin.bck. >>     >> >  > > >$ BACKUP USER$DISK:[HTTPD.AAPCO.BIN...] AAPCOBIN.BCK/SAVE_SET > D >plus any other qualifiers you normally use. No special qualifier is/ >needed to save the owner -- it's already done.  >  >  >    >  >>Also,  >>A >>once I copy the *.bck file to the new node and it is sitting in . >>web$disk:[httpd.aapco] what command do I useH >>to restore it with replace capability to web$disk:[httpd.aapco.bin...]; >>and keep the original owner of the directories and files?  >>     >> >  > 7 >$ BACKUP AAPCOBIN.BCK [*...]/BY_OWNER=ORIGINAL/REPLACE  >  > G >With VMS BACKUP, [*...] is "secret code" that means "restore the files B >to the same directories they were saved from, but on the "currentG >device"". Be advised that this will restore *all* files present in the G >save set. If the save set had files from [HTTPD.AAPCO...], e.g., you'd = >have to add /SELECT=[HTTPD.AAPCO.BIN...] after the save-set.  > F >You *do* need a qualifier to restore to the original owner except forA >/image, /incremental, and /physical restore operations. And that & >qualifier is the /BY_OWNER qualifier. > H >Please read the documentation for BACKUP in the System Utilities manual >for more information. >  >  >    >  >>Thank you in advance.  >>     >> >  >AEF >  >    >    ------------------------------    Date: 29 Jun 2006 10:43:45 -0500 From: briggs@encompasserve.org Subject: Re: VMS Backup 3 Message-ID: <Bvj6N641TJNp@eisner.encompasserve.org>   i In article <1151594237.760778.313930@75g2000cwc.googlegroups.com>, "AEF" <spamsink2001@yahoo.com> writes:  > Chuck Aaron wrote:K >> If I am sitting in a directory called user$disk:[httpd.aapco] and I want - >> to backup everything in the .bin directory I >> tree under aapco, what command would I use to insure the bin directory + >> and all directories and files underneath ' >> bin maintained their existing owner?  >>K >> For instance, let's say I wanted to use the backup command to save it to ! >> a saveset called aapcobin.bck.  >  > ? > $ BACKUP USER$DISK:[HTTPD.AAPCO.BIN...] AAPCOBIN.BCK/SAVE_SET  > E > plus any other qualifiers you normally use. No special qualifier is 0 > needed to save the owner -- it's already done. >  >  >> Also, >>B >> once I copy the *.bck file to the new node and it is sitting in/ >> web$disk:[httpd.aapco] what command do I use I >> to restore it with replace capability to web$disk:[httpd.aapco.bin...] < >> and keep the original owner of the directories and files? >  > 8 > $ BACKUP AAPCOBIN.BCK [*...]/BY_OWNER=ORIGINAL/REPLACE  B $ BACKUP AAPCOBIN.BCK /SAVE_SET [*...] /BY_OWNER=ORIGINAL /REPLACE                       ^^^^^^^^^ H Otherwise you simply get a disk-to-disk file copy of the saveset itself.   ------------------------------    Date: 29 Jun 2006 08:46:25 -0700$ From: "AEF" <spamsink2001@yahoo.com> Subject: Re: VMS Backup C Message-ID: <1151595985.820638.152890@j72g2000cwa.googlegroups.com>    bri...@encompasserve.org wrote: k > In article <1151594237.760778.313930@75g2000cwc.googlegroups.com>, "AEF" <spamsink2001@yahoo.com> writes:  > > Chuck Aaron wrote:M > >> If I am sitting in a directory called user$disk:[httpd.aapco] and I want / > >> to backup everything in the .bin directory K > >> tree under aapco, what command would I use to insure the bin directory - > >> and all directories and files underneath ) > >> bin maintained their existing owner?  > >>M > >> For instance, let's say I wanted to use the backup command to save it to # > >> a saveset called aapcobin.bck.  > >  > > A > > $ BACKUP USER$DISK:[HTTPD.AAPCO.BIN...] AAPCOBIN.BCK/SAVE_SET  > > G > > plus any other qualifiers you normally use. No special qualifier is 2 > > needed to save the owner -- it's already done. > >  > > 
 > >> Also, > >>D > >> once I copy the *.bck file to the new node and it is sitting in1 > >> web$disk:[httpd.aapco] what command do I use K > >> to restore it with replace capability to web$disk:[httpd.aapco.bin...] > > >> and keep the original owner of the directories and files? > >  > > : > > $ BACKUP AAPCOBIN.BCK [*...]/BY_OWNER=ORIGINAL/REPLACE > D > $ BACKUP AAPCOBIN.BCK /SAVE_SET [*...] /BY_OWNER=ORIGINAL /REPLACE! >                       ^^^^^^^^^ J > Otherwise you simply get a disk-to-disk file copy of the saveset itself.    1 Yes, you are right. Thanks for catching my error.    AEF    ------------------------------    Date: 29 Jun 2006 08:41:12 -0700$ From: "AEF" <spamsink2001@yahoo.com> Subject: Re: VMS Backup C Message-ID: <1151595672.033259.251810@i40g2000cwc.googlegroups.com>    Chuck Aaron wrote:F > SYSTEM> BACKUP/confirm AAPCOBIN.BCK [*...]/BY_OWNER=ORIGINAL/REPLACE7 > WEB$DISK:[HTTPD.AAPCO]AAPCOBIN.BCK;1, copy? (Y or N):  > G > with the above statement, it is only going to put the first file into # > web$disk:[httpd.aapco] instead of . > web$disk:[httpd.aapco.bin] where it belongs. >  > Chuck     Oops! I forgot to add /SAVE_SET.   Try   ? $ BACKUP AAPCOBIN.BCK/SAVE_SET [*...]/BY_OWNER=ORIGINAL/REPLACE      >  > AEF wrote: >  > >Chuck Aaron wrote:  > >  > > L > >>If I am sitting in a directory called user$disk:[httpd.aapco] and I want. > >>to backup everything in the .bin directoryJ > >>tree under aapco, what command would I use to insure the bin directory, > >>and all directories and files underneath( > >>bin maintained their existing owner? > >>L > >>For instance, let's say I wanted to use the backup command to save it to" > >>a saveset called aapcobin.bck. > >> > >> > >  > > @ > >$ BACKUP USER$DISK:[HTTPD.AAPCO.BIN...] AAPCOBIN.BCK/SAVE_SET     This could also be    ( $ BACKUP [.BIN...] AAPCOBIN.BCK/SAVE_SET  7 sinc you are already at USER$DISK:[HTTPD.AAPCO]. Sorry.      [...]    ------------------------------    Date: 29 Jun 2006 08:41:45 -0700$ From: "AEF" <spamsink2001@yahoo.com> Subject: Re: VMS Backup B Message-ID: <1151595705.746232.22480@d56g2000cwd.googlegroups.com>   Ian Miller wrote: 	 > I think  > & > $ BACKUP [.bin...] aapcobin.bck/save   Yes, of course, just [.bin...].    > 7 > $ BACKUP aapcobin.bck/save [.bin...]/OWN=ORIG/REPLACE   = I don't think this will do what the OP wanted. It will create    [.bin.httpd.aapco.bin...]   	 instead.     AEF    ------------------------------  % Date: Thu, 29 Jun 2006 08:55:43 -0500 - From: Hunter Goatley <goathunter@goatley.com> P Subject: Re: zipping large files (was Re: Info-ZIP's Zip V2.32 is now available)9 Message-ID: <lZQog.16303$7K2.4597@bignews2.bellsouth.net>    Steven M. Schweda wrote: > 0 >>  I was under the impression that you were theH >> "official" maintainer of zip/unzip for VMS, hut now, MR Goatley comesI >> out with Info-Zip which leads me to believe there are different source  >> trees/products around.  > H >    I'm just a (hard[ly]-working) flunky (who got into this whole thingD > by accident -- See "Zip -VV").  Mr. Goatley has been providing VMSH > binary distributions for a long time, and I'm certainly willing to letF > him continue to do so.  (How could I stop him?  And why would I?)  IF > believe that he added some IA64 support to the builders which he wasJ > distributing before the official source kits included that feature.  The) > official personel list is available at:  > ' >       http://www.info-zip.org/#People  > C > Pointers to official source and binary kits are available nearby.  >   E And many thanks to Steven for stepping in when he did. I've been part F of the Info-ZIP team since '91 or so (I did the original port to AlphaD and the original CLI version of Zip/UnZip), and as Steven said, I've@ been distributing VMS binaries via my freeware archive all theseE years.  Over the past few years, I've had less time to devote to the  B project. Fortunately for all of us, Steven joined the team and has@ done a lot of great work on Zip and UnZip, both the VMS-specific features and general features.   --     Hunter ------9 Hunter Goatley, Process Software, http://www.process.com/ B PreciseMail Anti-Spam Gateway for OpenVMS, Tru64, Solaris, & Linux9 goathunter@goatley.com     http://www.goatley.com/hunter/    ------------------------------  # Date: Thu, 29 Jun 2006 14:54:48 GMT , From: Hoff Hoffman <hoff-remove-this@hp.com>P Subject: Re: zipping large files (was Re: Info-ZIP's Zip V2.32 is now available), Message-ID: <YQRog.2$NH4.1@news.cpqcorp.net>   Craig A. Berry wrote: ? > In article <44a30fba$0$67258$157c6196@dreader2.cybercity.dk>, & >  "Dr. Dweeb" <spam@dweeb.net> wrote: > K >> Since large file support is still a ways off (zip specification ??), is  L >> there a VMS compression tool ala ZIP that is stable and usable for large  >> (30GB+) files ?  D    I have a gzip port (1.3.3?) that has been in use for OpenVMS I64 D DVDs, and we regularly use it for files beyond what zip can manage. @ (IIRC, current versions of zip tend to tip over at or near 2GB.)  I > I have used bzip2 to zip large files (~12GB or so) on other platforms,  F > and I know assorted bzip2 versions are available on VMS.  It should I > work fine, but I can't say I've actually done the exact thing you need  D > to do.  One advantage of bzip2 is that it gets better compression & > ratios than anything else I've seen.  H    I ported a version of bzip2 a while back, it wasn't a big deal.  The 5 version I ported went out on the Freeware.  V6 or V7?   G    unrar is also around, a port was just announced over at OpenVMS.Org.    ------------------------------   End of INFO-VAX 2006.359 ************************                                                                                                                                                                                                                                                                                                                                                      7I;&.	X@uqgƟBD;+r.aͥ}W 1 w:r}	x="XK<}5tZw0"$/K$&YŤ!e{FXP'1.I
[I4(r@I|d)^w#pE1~넋2SOm%vm]/8|+F+ʆ̓Bx0sm(&4ƃn]v!|J3sTVj#xv06Go׏!Y5BBm.=V0esdg:TF4a5gjtJ$V1}{Tsan&d%.w`,\Z0jH킟ṽ(EFJ+I;SWS#&bGuD½EIt̶^k
7Ҭ	3"'b>Sl0V%эJD2SմpSҀK,sּ,܁)@͸Q6:GZP;V'f:ccؽ lwm"L274#Y%a1-'x	Nkd]Y@ZI!evAB-71eqfĠHz:"<j("[&pfD;K+v~3,
6{t1:mgVx5P Y$8i4s`<)6D׋:tteXEz:uh.n3WAoG<p}rh9VPIAuٯ1bgJ
9N_oDVĮ׵5F{po2QQ3
6$7X]q82j~v&37b!T̝i$FVxMrcG	rᔽ*V˗V¬2?'E9#cF[eJNl;+Ҭ2((
0Š][EPG	"qgmxINLݾ@CFiy֮0G|+F%pJɡfQy^]vchN2	*&"nb7~=s
Fv#_Ɏ_<kޢR4&ؔp?4O??e	=3+[ b$c0fK;3yNw|㱹#a6]bp~18/F+18h@-iWfS4؜ZEI:
&FoFVI;G(D
U)SN-oCYsR`/qJ[K
%
9;lB&woidM|h<
NIi{qh
;B`k9uJDdb"a\+ۧf}cDQn)DOu#a/PY$8ת.@YSy#	g| [ԊkY
i6*BeܐK4H%iT0­=>Gʑ]Fal[G뫁
s3??ܸKARؐJ^z*IMAxU5A0Ӧ >XlsƧgM
i R8^j8Do/	.ސZd]CA#[ehA[y*R 7I/jB!8
rT8I7V,r|$6MEQ%HKE_qȄJzqKd	GnGn,'ǍuC@,*/+%@ueËwyWSf[n\{ӦQ35y3t;Wcӊf|;IC
WTXsO盅8ڄ~
A
d90[6N眚MhU7uPIvdLYĄ7Pc[~9;4SP#d|[w<~kiΩ