Hi Craig! Long time no talk.
> >I'm trying to install GD 2.35 under HP's Perl 5.8.6, running
> >under OpenVMS ALpha V7.3-2 .
> >I have built all the libraries I need (zlib 1.2.3, libpng 1.2.16,
> >libjpeg 6B, gd 2.0.34), and after some Makefile.PL tweaking, the
> >MMK run completes without complaints (see log below).
> >
> >BUT: The resulting PL_GD.EXE shareable does not export any
> >universal symbols except BOOT_GD ?!
>
> How universal do they need to be? I don't think the Perl extension
> building process expects anything outside the extension to call into
> it, so I think what you're seeing is normal.
Okay. Seems like a false assumption on my part. What I saw is a
test failure in t/gd.t after what seemed to be a seamless build:
$ mmk test
MCR $1$dga100:[tools.perl5_8_6]perl.exe "-MExtUtils::Command::MM"
"-e" "test_harness(0, '[.blib.lib]', '[.blib.arch]')" t/*.t
t/gd..........
Use of uninitialized value in concatenation (.) or string at t/gd.t line 6.
Regression tests require one or more of PNG, GIF or JPEG support in libgd. at
t/gd.t line 29.
%SYSTEM-F-NOLOGNAM, no logical name match
dubious
Test returned status 444 (wstat 1024, 0x400)
(VMS status is 444)
...
The message "Regression tests require..." stems from the env var GDIMAGETYPE
not being defined and neither newFromPng nor newFromJpeg nor newFromGif being
recognized by GD::Image (or UNIVERSAL) as valid class methods.
Which led me to the conclusion that this failure was soimthing so basic
that it had something to do with exported symbols... Sorry.
But looking at the other error message (t/gd.t line 6), there seems
more that's not quite right:
$ perl "-MFindBin=$Bin" -e "print ""$Bin\n"""
DISK$DATA:[TEMP.WEB_NEU.SRC.MODULE.GD-2_35]
$
> I just ran
>
> $ search [.ext...]*.opt symbol_vector
>
> on a recently built source directory, and none of the core extensions
> exports anything except the boot symbol.
>
> >I *think* I traced it to Makefile.PL not properly setting
> >the DL_FUNCS option to WriteMakefile, but if I try to correct
> >that manually, the resulting MMK action line is too long to
> >be executed.
> >
> >Help?
>
>
> I'm surprised you ran into command-line length problems on 7.3-2 as I
> thought that was where the buffers got a big boost. But I'm puzzled
> about what the real problem is. As you say, MMK ran to completion
> without any serious errors, so what exactly is the problem? If you
> were hoping to build a combined shareable image that combines a
> number of different libraries, then that may involve some real work
> as Perl extensions don't usually export everything that they import.
No, nothing quite so exotic. I just need GD to work properly.
Thanks for your comments.
cu,
Martin
--
| Martin Vorlaender | OpenVMS rules!
Microsoft isn't the Borg: | work: [EMAIL PROTECTED]
the Borg have proper | http://www.pdv-systeme.de/users/martinv/
networking. | home: [EMAIL PROTECTED]
