|
|
Subject: Re: my first builds with cygwin 1.5.3-1 (blead,
5.8.1, 5.6.2) - msg#01264
Hallo Yitzchak,
Am Montag, 22. September 2003 um 06:00 schriebst du:
> On Sun, Sep 21, 2003 at 03:12:00PM -0700, Yitzchak Scott-Thoennes
> <sthoenna@xxxxxxx> wrote:
>> >
>> > Wow, the build of 5.6.2@21256 on 1.3.22 fails completly for me, havn't
>> > tried with 1.5.4 yet. I crafted a patch today, will post it tomorrow
>> > when I saw the test results.
>>
>> I didn't see this patch.
> Yes I did, I just forgot about it after reading it. I probably didn't
> have the same problems because I think I accidentally ended up with a
> perlld from a newer perl.
Usually, ld2 is copied to your path (/usr/bin) which calls perlld in
the build directory. Iff you have something around, it should be an
old ld2 wrapper e.g. in /usr/local/bin. This was probably also the
problem why it failed for me.
> Is there a good explanation anywhere of why perlld is needed?
Not that I'm aware of. It was introduced because cygwin used dllwrap
to create dll's which is not a single pass step as creating them with
gcc -shared -o perl.dll ...
I would like to get rid of it as soon as possible. Since gcc
understands now the -shared flag building shared libs is pretty much
the same as on e.g. linux besides that we want an import library and
a shared library (libperl.dll.a & libperl.dll) and also a static
libperl.a for miniperl.
So we could use gcc as linker instead of the wrapper ld2, will be just
a little tricky to get all the ldflags at the right place.
Gerrit
--
=^..^=
Thread at a glance:
Previous Message by Date:
Re: my first builds with cygwin 1.5.3-1 (blead, 5.8.1, 5.6.2)
Hallo Yitzchak,
On Sep 22, 2003 at 00:12 Yitzchak Scott-Thoennes wrote:
> On Wed, Sep 17, 2003 at 11:33:17PM +0200, "Gerrit P. Haase"
> <gp@xxxxxxxxxxxxxxx> wrote:
>> Am Mittwoch, 17. September 2003 um 20:30 schriebst du:
>> > cygwin now at 1.5.3.
>>
>> > The ping failures are hangs...they happen even with 5.8.0 and are
>> > probably something messed up on my end.
>>
>> > bleadperl@21260:
>>
>> > Failed Test Stat Wstat Total Fail Failed List of
>> > Failed
>> > -------------------------------------------------------------------------------
>> > ./ext/Socket/t/socketpair.t 1 256 45 1 2.22% 14
>> > ./lib/Net/Ping/t/510_ping_udp.t 0 15 2 2 100.00% 2
>> > op/lfs.t 17 4 23.53% 6 8 10 12
>> > op/sysio.t 39 1 2.56% 39
>>
>> > perl-5.8.x@21251:
>>
>> > Failed Test Stat Wstat Total Fail Failed List of
>> > Failed
>> > -------------------------------------------------------------------------------
>> > ./ext/Socket/t/socketpair.t 1 256 45 1 2.22% 14
>> > ./lib/Net/Ping/t/510_ping_udp.t 0 15 2 2 100.00% 2
>> > io/tell.t 28 1 3.57% 28
>> > op/lfs.t 17 5 29.41% 6 8 10 12
>> > 17
>> > op/sysio.t 39 1 2.56% 39
>>
>> I see the sysio.t and the tell.t too, but not the others (and I'm on
>> cygwin 1.5.4).
> Everything is the same with 1.5.4:
> perl-5.6.2/.patch:21281
> perl-5.8.x/.patch:21300
> perl/.patch:21299
> Now I have to try 1.5.5 w/gcc 3.3 :)
It is gcc-3.3.1, gcc-3.3 was not released for cygwin.
> Do you know why tell.t is failing in 5.8.1 but not bleadperl?
It is a perlio test. I removed Cygwin from the list of Dosish
platforms at the beginning of the test, but the last test is failing
also then. I have no idea why. Basically, some bytes are written to
a file and then a tell() on the file should return the count of bytes,
instead of 6 tell() counts 0 bytes.
> Here's the op/lfs.t output from perl-5.8.x in case it helps anyone:
The lfs.t is always skipped for me, "no sparse files".
> --------------------------------------------------
> # checking whether we have sparse files...
[...]
>> > perl-5.6.2@xxxxxxxxx:
>>
>> > Failed Test Stat Wstat Total Fail Failed List of Failed
>> > -------------------------------------------------------------------------------
>> > lib/io_xs.t 4 2 50.00% 2-3
>> > op/lfs.t 17 5 29.41% 6 8 10 12 17
>> > op/sysio.t 39 1 2.56% 39
I see the sysio.t failing but not the others, lfs.t is skipped as
mentioned above. Could you post the output of the io_xs.t, please?
>> Wow, the build of 5.6.2@21256 on 1.3.22 fails completly for me, havn't
>> tried with 1.5.4 yet. I crafted a patch today, will post it tomorrow
>> when I saw the test results.
> I didn't see this patch.
Hmmm, unfortunately the mailer I'm using to read perl5-porters is a
newsreader and it doesn't produce message IDs, these were the headers
of the mail where the patch was attached. However, if it builds for
you, but not for me then I guess my setup is somewhat broken and the
patch it is probably not needed, but anyway it is no longer supported
on Cygwin to use dlltool and dllwrap to build shared libraries since
gcc does the right thing when passing -shared.
From: Gerrit P. Haase <gerrit@xxxxxxxxxxxxxxx>
To: perl5-porters@xxxxxxxx
Subject: Re: perl562@21256
Date: Thu, 18 Sep 2003 12:08:40 +0200
Organization: Esse keine toten Tiere
Reply-To: gerrit@xxxxxxxxxxxxxxx
References: <20030917004229.198b0bc8.rgarciasuarez@xxxxxxx>
In-Reply-To: <20030917004229.198b0bc8.rgarciasuarez@xxxxxxx>
X-Mailer: Forte Agent 1.91/32.564
--
=^..^=
Next Message by Date:
[perl #23893] Re: How can I comment out a large block of perl code?
# New Ticket Created by Dan Jacobson
# Please include the string: [perl #23893]
# in the subject line of all future correspondence about this issue.
# <URL: http://rt.perl.org/rt2/Ticket/Display.html?id=23893 >
The following message is a courtesy copy of an article
that has been posted to comp.lang.perl.misc as well.
D> my bugs just go into a black hole.
A> What do you mean? Some Cabale?
Well, ok, I suppose I was just used to Debian's auto etc. acknowledgments.
Anyway: Dear CC'd perlbug: see the current thread
http://groups.google.com/groups?threadm=bkfl3b%24b94%241%40mamenchi.zrz.TU-Berlin.DE
for mention of perlpod undocumented features etc.
Previous Message by Thread:
Re: my first builds with cygwin 1.5.3-1 (blead, 5.8.1, 5.6.2)
On Sun, Sep 21, 2003 at 03:12:00PM -0700, Yitzchak Scott-Thoennes
<sthoenna@xxxxxxx> wrote:
> >
> > Wow, the build of 5.6.2@21256 on 1.3.22 fails completly for me, havn't
> > tried with 1.5.4 yet. I crafted a patch today, will post it tomorrow
> > when I saw the test results.
>
> I didn't see this patch.
Yes I did, I just forgot about it after reading it. I probably didn't
have the same problems because I think I accidentally ended up with a
perlld from a newer perl.
Is there a good explanation anywhere of why perlld is needed?
Next Message by Thread:
Re: my first builds with cygwin 1.5.3-1 (blead, 5.8.1, 5.6.2)
Hallo Yitzchak,
On Sep 22, 2003 at 00:12 Yitzchak Scott-Thoennes wrote:
> On Wed, Sep 17, 2003 at 11:33:17PM +0200, "Gerrit P. Haase"
> <gp@xxxxxxxxxxxxxxx> wrote:
>> Am Mittwoch, 17. September 2003 um 20:30 schriebst du:
>> > cygwin now at 1.5.3.
>>
>> > The ping failures are hangs...they happen even with 5.8.0 and are
>> > probably something messed up on my end.
>>
>> > bleadperl@21260:
>>
>> > Failed Test Stat Wstat Total Fail Failed List of
>> > Failed
>> > -------------------------------------------------------------------------------
>> > ./ext/Socket/t/socketpair.t 1 256 45 1 2.22% 14
>> > ./lib/Net/Ping/t/510_ping_udp.t 0 15 2 2 100.00% 2
>> > op/lfs.t 17 4 23.53% 6 8 10 12
>> > op/sysio.t 39 1 2.56% 39
>>
>> > perl-5.8.x@21251:
>>
>> > Failed Test Stat Wstat Total Fail Failed List of
>> > Failed
>> > -------------------------------------------------------------------------------
>> > ./ext/Socket/t/socketpair.t 1 256 45 1 2.22% 14
>> > ./lib/Net/Ping/t/510_ping_udp.t 0 15 2 2 100.00% 2
>> > io/tell.t 28 1 3.57% 28
>> > op/lfs.t 17 5 29.41% 6 8 10 12
>> > 17
>> > op/sysio.t 39 1 2.56% 39
>>
>> I see the sysio.t and the tell.t too, but not the others (and I'm on
>> cygwin 1.5.4).
> Everything is the same with 1.5.4:
> perl-5.6.2/.patch:21281
> perl-5.8.x/.patch:21300
> perl/.patch:21299
> Now I have to try 1.5.5 w/gcc 3.3 :)
It is gcc-3.3.1, gcc-3.3 was not released for cygwin.
> Do you know why tell.t is failing in 5.8.1 but not bleadperl?
It is a perlio test. I removed Cygwin from the list of Dosish
platforms at the beginning of the test, but the last test is failing
also then. I have no idea why. Basically, some bytes are written to
a file and then a tell() on the file should return the count of bytes,
instead of 6 tell() counts 0 bytes.
> Here's the op/lfs.t output from perl-5.8.x in case it helps anyone:
The lfs.t is always skipped for me, "no sparse files".
> --------------------------------------------------
> # checking whether we have sparse files...
[...]
>> > perl-5.6.2@xxxxxxxxx:
>>
>> > Failed Test Stat Wstat Total Fail Failed List of Failed
>> > -------------------------------------------------------------------------------
>> > lib/io_xs.t 4 2 50.00% 2-3
>> > op/lfs.t 17 5 29.41% 6 8 10 12 17
>> > op/sysio.t 39 1 2.56% 39
I see the sysio.t failing but not the others, lfs.t is skipped as
mentioned above. Could you post the output of the io_xs.t, please?
>> Wow, the build of 5.6.2@21256 on 1.3.22 fails completly for me, havn't
>> tried with 1.5.4 yet. I crafted a patch today, will post it tomorrow
>> when I saw the test results.
> I didn't see this patch.
Hmmm, unfortunately the mailer I'm using to read perl5-porters is a
newsreader and it doesn't produce message IDs, these were the headers
of the mail where the patch was attached. However, if it builds for
you, but not for me then I guess my setup is somewhat broken and the
patch it is probably not needed, but anyway it is no longer supported
on Cygwin to use dlltool and dllwrap to build shared libraries since
gcc does the right thing when passing -shared.
From: Gerrit P. Haase <gerrit@xxxxxxxxxxxxxxx>
To: perl5-porters@xxxxxxxx
Subject: Re: perl562@21256
Date: Thu, 18 Sep 2003 12:08:40 +0200
Organization: Esse keine toten Tiere
Reply-To: gerrit@xxxxxxxxxxxxxxx
References: <20030917004229.198b0bc8.rgarciasuarez@xxxxxxx>
In-Reply-To: <20030917004229.198b0bc8.rgarciasuarez@xxxxxxx>
X-Mailer: Forte Agent 1.91/32.564
--
=^..^=
|
|