Craig A. Berry [EMAIL PROTECTED] wrote:
At 10:47 AM 3/11/2002 -0800, Mike Scott wrote:
my specific problem is with seek(). My goal? To read the last few
lines of a text file. I've tried seek(FCB, -200, 2) which should
redirect me to 200 bytes before eof.
It should if it's a stream file.
Craig A. Berry [EMAIL PROTECTED] wrote:
You mean an equivalent to the F$FILE_ATTRIBUTES lexical function?
Yeah. Of course, one could always invoke _that_ from Perl:
`write sys\$output f\$file_attributes ($file_name, RFM)`
And no, that's not an implied slam on manual-reading abilities - it's
Michael G Schwern [EMAIL PROTECTED] wrote:
ExtUtils::MakeMaker added this file to Perl:
lib/ExtUtils/t/Big-Fat-Dummy/lib/Big/Fat/Dummy.pm
is that too long for old VMSes?
Technically, no - but it's right at the brink, and I hope you won't, since
I'm stuck with one of the old VMSes.
$
Dan Sugalski [EMAIL PROTECTED] wrote:
The 8 level limit applies from the root of the device you're
using, but if that device is a translated concealed logical it
still works. (You can get 15 levels deep this way)
Well, Perl works, but other things (like VMS BACKUP) get serious
indigestion
Michael G Schwern [EMAIL PROTECTED] wrote:
On Mon, Apr 01, 2002 at 09:17:41AM -0500, Thomas R Wyant_III wrote:
Dan Sugalski [EMAIL PROTECTED] wrote:
The 8 level limit applies from the root of the device you're
using, but if that device is a translated concealed logical it
still
Craig A. Berry [EMAIL PROTECTED] wrote:
Well, here's how to pretend that the current directory is the root of
a volume. Having the test do something like this before running
Makefile.PL is probably what we'll have to do:
[DCL script example omitted]
I had my doubts about this approach in
[EMAIL PROTECTED] wrote:
The 8 levels before translation and 15 after translation are NOT
a problem for VMS they are a limitation imposed by the RMS path
cache.
This statement is technically correct. I'm not sure how we can apply it to
the problem at hand, though. If you built Perl under
[EMAIL PROTECTED] wrote:
They [the RMS libraries] are in the Executive, not linked to
your process, so a VMS upgrade is needed.
ACK. I was just trying to make the point that whether the directory depth
limit is core VMS or RMS, there still seems to be no practical way to deal
with it other
All,
I'm having a problem getting correct times from the stat () function under
Perl 5.6.1/OpenVMS/AXP 7.1-1H (details appended). I suspect a configuration
problem on the VMS system in question, but am out of places to look. A
quick demo:
$ perl -e print scalar localtime
Mon Jun 3 14:25:46
Craig,
Based on my reading of Compaq's web site, we're doing the time change
correctly, though sloppily. I did find an article on Issues with mktime
and stat function, though the circumstances of their problem were somewhat
different than mine. Based on this information, I duplicated the
Brian Tillman [EMAIL PROTECTED] wrote:
What I'm greatful for is that the perl maintainers don't all say, this
VMS-specific crap isn't worth our time. Let's just drop that platform.
Amen!
Although there is certainly some give-and-take among users of the various
OSes, the Perl community
All,
Up until now, I have mostly ignored tools like perldoc and pod2* under VMS.
I'm trying to get them going now, and have encountered an anomaly using
Craig Berry's 5.6.1 kit under VMS 7.1-1H2:
The anomaly (non-fatal!) is that perldoc generates the error
Use of uninitialized value in
All -
I have been in contact with Ohio State University about the WEBPERL.EXE
component of their DECThreads web server.
Not only have files moved in the Perl directory tree, but the old version
of WEBPERL doesn't support the current CGI.pm (they overrode %ENV to get
the CGI environment
Craig A. Berry [EMAIL PROTECTED] wrote:
Where Perl can really slaughter DCL is not surprisingly in text
processing. If you have procedures that make heavy use of F$ELEMENT,
F$LOCATE, or do text substitutions and such then hundreds of lines
of DCL can sometimes be reduced to a handful of
Haining,
What I generally do is to build the module in place. Typically that just
means expanding the tar file, which typically creates a subdirectory
anyway, SETting DEFAULT to that subdirectory, and then
$ perl makefile.pl
$ mmk
$ mmk test ! If there is one
$ ! Privilege may be needed for
Craig A. Berry [EMAIL PROTECTED] @ DUPONT_MHUB wrote:
I'm no Apache guru so those are just guesses as to what might be your
problem, but this sounds like a side effect of the fact that there is
no execute bit on files and no shebang line processing on VMS.
I'm not either, but in the absence
Patrick,
VMS is not Unix, but you _may_ be able to get your Unix syntax to work if
you hold your tongue correctly.
At some point in the history of VMS, a command called pipe was
introduced, which appears to implement a lot of Unix I/O redirection
functionality, including pipes and redirection.
Craig A. Berry [EMAIL PROTECTED] @ DUPONT_MHUB wrote:
5.8.0 is the current version on all supported platforms. The current
stable version is always available from
http://www.cpan.org/src/stable.tar.gz
Any chance of an easy-to-install package for 5.8.0, similar to the 5.6.1
package at
Patrick,
Why not just use 21? No, it's not VMS syntax. _NO_ Unix-style redirection
is VMS syntax. So the VMS version of Perl provides it in so far as it is
able. Sample run appended, with node names removed to protect the guilty.
Interestingly, I have worked with VMS for quite a while, and I
Brian Tillman [EMAIL PROTECTED] wrote:
Tom Wyant wrote:
Why not just use 21? No, it's not VMS syntax. _NO_ Unix-style
redirection
is VMS syntax.
According to HELP PIPE PARAMETERS, Unix-style redirection is allowed.
You're right; the PIPE command allows a lot of Unixy-style stuff. But
Martin Vorlaender [EMAIL PROTECTED] wrote:
You prefix the perl script with
$ perl 'f$env(procedure)' ''p1' ''p2' ''p3' ''p4' ''p5'
''p6'
''p7' ''p8'!
$ exit++ + ++$status!=0 and $exit=$status=undef; while($#ARGV != -1 and
$ARGV[$#ARGV] eq ''){pop @ARGV;}
eval 'exec -S $0 ${1+$@}'
Brad Hughes brad/[EMAIL PROTECTED] wrote:
Brian Tillman wrote:
VMS IS NOT a Unix descendent.
Certainly Unix predates VMS. And Multics predated Unix. The
TOPSes predated VMS, but I'm not sure you could say VMS is
strictly descended from TOPS. Did Multics and TOPS share a
common
Henderson, Jordan (Contractor) (DAASC) [EMAIL PROTECTED] wrote:
RSX came about just about the same time as Unix, and I would be surprised
if anything from Unix made it into the RSX design or implementation.
Unix people like to harken back to it's origins in 1969, but those first
few years
Possible flamebait follows. Read at own risk.
Carl Friedberg [EMAIL PROTECTED] wrote:
(5) PDP-8 (12 bit?). I don't remember much of anything about that, but
it morphed into a word processor :-)
You're right, it was 12-bit. About 10 years ago we were working in Spain
for a year. We received a
Keirnan, Paul [EMAIL PROTECTED] wrote:
Tom wrote:
From: Thomas R Wyant_III [mailto:[EMAIL PROTECTED]
Question: Why is the VMS image activator so slow?
Answer: Same reason TKB is.
I was sent this years ago...
(Gripping drama of life in the mean streets of VMS excised)
As I recall
Robert Atkinson wrote:
Remember, you're talking to someone who's used to programming in DCL,
which
will almost always tell you when it encounters problems. Now I've begun
using
this 'new' language Perl, I expect it to act similarly. When it doesn't,
I
need to find out..
- Why?
- How do I
Craig Berry wrote:
Excellent advice, but often harder to follow than it should be. Think of
trying to explain to a VMS
newbie why the RMS system services are not to be found in the System
Services Reference
Manual, the CVT$ routines are in the RTL Library (LIB$) even though
they don't begin
Robert,
Jump-starting yourself with knowledge of something similar is a tried and
true technique. It's a little harder to make it work with Perl, though,
since Perl is so different. Larry Wall in the Camel Book compares parts of
it to Basic, parts to Fortran, parts to C, and so on. Before I got
Jim,
It seems to me that there may be a nomenclature problem here. What _I_ call
a text file with binary data is a binary file. It might be as simple as
inserting a binmode ($fh) after the vmsopen. If that doesn't work, look
at the other vmsopen arguments.
Kicking it up a level, are you just
[EMAIL PROTECTED] wrote:
Since perl 5.6.0 binmode is a noop as it is for Unix.
Interesting. $ perldoc -f binmode and $perldoc perlvms seem to me to
say otherwise. But under 5.6.1 it certainly behaves like a noop. On output
you get stream_lf files, so it _would_ be a no-op. It seems to just
Craig A. Berry [EMAIL PROTECTED] wrote on 08/23/2004 11:08:38 PM:
Hi Folks,
This isn't the latest greatest since 5.8.5 is out now, but these are
much more recent than other binary kits available (such as the now
ancient 5.6.1 still distributed by HP). Here's what I've got:
Binary kits for
Craig A. Berry [EMAIL PROTECTED] wrote on 08/25/2004 11:13:57 PM:
snip! /
I tried this [the easy-to-build source kit] yesterday, doing everything
under my own username, with all
privileges except NETMBX and TMPMBX turned off. The environment is:
VMS: 7.1-1H2
CC: DEC C V5.6-003
TCP/IP:
Sam Ravindhran [EMAIL PROTECTED] wrote on 10/24/2004 09:06:49 PM:
The following lines of code taken from Unix fails for
me on VMS:
--
if(open(EXEC_WRITER, |$executable $filename))
{
Snip!
I get an error indicating that the character is
invalid in a command
All,
I have a need to spawn a command from a Perl script, time it out, and kill
it if the timeout occurs. Under Perl 5.6.1, something like the following
worked:
$| = 1;
my $cmd = whatever comand I want to issue;
my $tmot = 30; # Timeout in seconds.
my $pipe;
my $pid = open ($pid, $cmd|) or
Craig A. Berry [EMAIL PROTECTED] wrote on 01/14/2005 09:32:40 AM:
snip /
Surely you want
my $pid = open ($pipe, $cmd|) or die;
Absolutely right. I noticed this later, but the posting had already lain
fallow for a bit, and I decided to correct myself when I gave a progress
report, if nobody
Craig A. Berry [EMAIL PROTECTED] wrote on 01/14/2005 12:45:54 PM:
snip /
The way to find out what you're configured with is:
$ perl -V:useperlio
useperlio='define';
Yep, when I do that, I get useperlio='define';
snip /
I read the perlipc documentation the same way you do, i.e., the
Craig A. Berry [EMAIL PROTECTED] wrote on 01/17/2005 12:24:20 AM:
snip /
Well, of course it's not that simple. What was I thinking? :-( A
bit more analysis shows that where it gets stuck is in the parent,
which is not doing one of our pipe $QIOs at AST level, but is instead
sitting there
Craig A. Berry [EMAIL PROTECTED] wrote on 01/17/2005 11:55:12 PM:
snip /
If I can figure out what the unixes are doing and if it's something
we can do without causing other problems, I may yet solve this. But
I may not, and almost certainly won't get to it quickly. So I'm
afraid you are
Richard Kelly [EMAIL PROTECTED] wrote on 02/07/2005 02:34:40 PM:
snip /
Another issue I will probably encounter in the future is accessing DCL
symbols, logical names, and lexicals ( I have seen some discussion of
lexicals but I have not seen a conclusion).
DCL symbols: see VMS::DCLsym
[EMAIL PROTECTED] wrote on 02/08/2005 11:17:58 AM:
snip /
VMS::Stat handles many f$file_attribute()s (in fact one of it's
regression
tests does extensive comparisons between DCL and perl extension results
for a test file). It does not handle time stamps well at all though
(CDT,
vs RDT vs
[EMAIL PROTECTED] wrote on 02/23/2005 08:10:49 PM:
[snip /]
I am sorry to say that I cannot help much here other than to point
out the examination of your fabdef.h header could help out, then pehaps
introducing some sensitivity to macros like:
Yes, I was pursuing conditionalizing on
Craig A. Berry [EMAIL PROTECTED] wrote on 02/24/2005 06:40:32 PM:
[snip /]
It looks like it's there as a macro on v7.3-1:
$ sea sys$common:[decc$lib.reference.sys$starlet_c]*.h fab$v_erl
**
SYS$COMMON:[DECC$LIB.REFERENCE.SYS$STARLET_C]FABDEF.H;3
Michael G Schwern [EMAIL PROTECTED] wrote on 03/13/2005 06:34:27 PM:
As mentioned here, I want to break up MakeMaker.
http://www.makemaker.org/wiki/index.cgi?ModulesForSale
There's a number of modules which are just utilities MakeMaker uses and
I
want them out in their own distribution.
Craig Berry [EMAIL PROTECTED] wrote on 03/14/2005 01:31:01 PM:
SNIP!
Yes, it should, but so should blib. I'm not convinced this is the
real issue, but not sure when I'll get a chance to dig into it myself.
Maybe that's why I didn't get anywhere pursuing it. But I remember not
being able
44 matches
Mail list logo
