Front page | perl.vmsperl |
Postings from June 2002
Brian Tillman wrote: !>I don't know that it's incorrect, though it could be done more simply like !> !>$ create 'file_2_find' !>$ set file/attribute=(RFM:STMLF,RAT:CR) 'file_2_find' !>$ open/append config 'file_2_find' ! !/ATTRIBUTE on SET FILE didn't exist prior to V6.0. Perhaps that's not a !concern. I am sorry for having set forth such a terse justification for the seemingly elaborate CREATE/FDL construct that I recently patched into perl. Here is a bit more detailed description: 1) vile 9.2 was shown to me to truncate the VFC bytes in a perl_setup.com written by an older configure.com, and indeed it omitted the leading "$" character that is needed for a properly constructed DCL procedure. While this should be considered a bug in vile, the DCL "@" command to execute perl_setup.com would not balk an executing either an RFM .eqs. "VAR" or an RFM .eqs. "STMLF" perl_setup.com. 2) certain other bits of software, among them the Perforce source code management system, and NFS clients such as those found on Solaris 2.8 are happier with RFM .eqs. "STMLF". In particular the VMS perforce client "sync" command will pull out a STMLF file from a Perforce depot. Adding a VFC or even a VAR RFM file to Perforce will autoconvert it to a Perforce binary file type that is not considered text hence is not considered amenable to perforce commands such as diff, diff2, or describe with the long output option. Here again this could be considered a bug in software that is not VMS native, but since "@" can accept STMLF it seemed preferable to have configure.com create perl_setup.com with the STMLF record format. Alas such a long winded explanation did not lend itself to a quick patch submission on p5p. That said I am not opposed to switching to VAR if you or anyone else can come forth with an inconvenience scenario that clearly demonstrates that it would be preferable to STMLF (fear not as I am familiar with the more succinct $CREATE $OPEN/APPEND idiom, the more verbose CREATE/FDL construct is actually used further up in the creation of config.h from config.local). Since I need to track changes to a perl_setup.com using Perforce I would need to use STMLF, but I can easily run a CONVERT/FDL over a VAR version of the file before checking it into my SCM system. Would that be your preference? If so why? Thanks. Peter Prymmer