Just to clarify one thing about fink's setup.
We store all XS modules in versioned subdirectories of /sw/lib/perl5.
The modules which are stored directly in /sw/lib/perl5 are the non-XS
modules, which should work the same across all perl versions.
(This wasn't always true, but we had a big cleanu
On Fri, Nov 14, 2003 at 09:22:47AM -0800, Randal L. Schwartz wrote:
> I am perhaps misremembering then. I recall seeing something on P5P
> that 5.8.2 focussed on being compatible with 5.8.0, and thus breaks
> with 5.8.1 in a few ways. It is still early in the morning here
> though.
You're right.
TheSin wrote:
acctually so the debs stay the same, fink needs to have a virtual pkg
for altivec and then have a lame-altivec version which would depend on
altivec and would could then use -faltivec...like i did with ffmpeg,
but with ffmpeg I forced it on, not really great but since there is n
though this should be Fink::PerlModule IMHO
On 14-Nov-03, at 9:17 AM, TheSin wrote:
if you start it I'll help, then i can add the auto depends stuff there
and pass it back to the dep engine..
On 14-Nov-03, at 8:29 AM, Michael G Schwern wrote:
Hell, it might even make sense to start a Fink::Per
Am Freitag, 14.11.03 um 16:43 Uhr schrieb TheSin:
good idea, I just think for others it should auto install the update
even if they don't know about it...fink-core is for project admins or
should i be on that list?
fink-core is not a secrete elite project admin group, rather it is a
group whic
Am Freitag, 14.11.03 um 17:16 Uhr schrieb TheSin:
couldn't you just use gnu ??
It's not on the normal GNU mirrors. It's alpha ware, so it is only on
the alpha.gnu.org FTP server.
Max
---
This SF. Net email is sponsored by: GoToMyPC
GoToMyPC
Am Freitag, 14.11.03 um 16:36 Uhr schrieb TheSin:
but what happens at compile time?? can altivec code be compiled on and
non altivec machine??
Of course. It only depends on the compiler, not the host machine,
whether altivec code can be compiled.
Max
what is up with finkcommander?? Does that make sense to anyone?
On 14-Nov-03, at 12:04 PM, Martyn Klassen wrote:
Mac OS X version: 10.3.1
December 2001 Developer Tools
XCode
gcc version: 3.3
---
This SF. Net email is sponsored by: GoToMyPC
Go
-BEGIN PGP SIGNED MESSAGE-
Hash: RIPEMD160
Max Horn wrote:
Am Freitag, 14.11.03 um 14:53 Uhr schrieb Darian Lanx:
-BEGIN PGP SIGNED MESSAGE-
Hash: RIPEMD160
David R. Morrison wrote:
I was looking at this last night, as Benjamin already posted our
conversation I can only add t
I'll put language to that effect then, and remove it when we've got a
working XFree86 again.
I've seen some fairly snippy messages about the docs, particularly
during the Panther transition, so I wanted to make this one
Panther-relevant.
On Friday, November 14, 2003, at 12:47 PM, Benjamin Reed
Alexander K. Hansen wrote:
I'm planning on updating the "Running X11" document to include
Panther/Apple X11 material. One additional issue came to my mind: Is
XFree86 still broken for Panther?
Yes. I've been talking with torrey trying to get it worked out, but so
far I haven't gotten a work
okay my bad then :)
On 14-Nov-03, at 10:26 AM, Benjamin Reed wrote:
TheSin wrote:
couldn't you just use gnu ??
Source: gnu
alpha.gnu.org != ftp.gnu.org :)
--
Benjamin Reed a.k.a. Ranger Rick -- http://ranger.befunk.com/
gpg: 6401 D02A A35F 55E9 D7DD 71C5 52EF A366 D3F6 65FE
"You CAN'T clean
I'm planning on updating the "Running X11" document to include
Panther/Apple X11 material. One additional issue came to my mind: Is
XFree86 still broken for Panther?
--
Alexander K. Hansen
Levitated Dipole Experiment
http://www.psfc.mit.edu/LDX
--
Michael G Schwern wrote:
On Fri, Nov 14, 2003 at 11:46:12AM -0500, Benjamin Reed wrote:
Was it an apple engineer that did this? Because Perl 5.8.1 final
out-of-the-box defaults to the same thing that's in Apple's perl.
5.8.1 doesn't default to using threads. Maybe you have a different box?
TheSin wrote:
couldn't you just use gnu ??
Source: gnu
alpha.gnu.org != ftp.gnu.org :)
--
Benjamin Reed a.k.a. Ranger Rick -- http://ranger.befunk.com/
gpg: 6401 D02A A35F 55E9 D7DD 71C5 52EF A366 D3F6 65FE
"You CAN'T clean the toilet, Neil, it'll lose all it's character!"
-- Vyvyan
pgp00
> "Michael" == Michael G Schwern <[EMAIL PROTECTED]> writes:
Michael> I think you're wrong about forward compatibility. The point
Michael> of 5.8.2 was to be compatible with 5.8.0 and 5.8.1. Its
Michael> 5.8.1 that introduced the binary backward compat glitch with
Michael> 5.8.0. Fortunatel
On Fri, Nov 14, 2003 at 11:46:12AM -0500, Benjamin Reed wrote:
> Was it an apple engineer that did this? Because Perl 5.8.1 final
> out-of-the-box defaults to the same thing that's in Apple's perl.
5.8.1 doesn't default to using threads. Maybe you have a different box?
--
Michael G Schwern
On Fri, Nov 14, 2003 at 08:27:37AM -0800, Randal L. Schwartz wrote:
> Not to get too confused, but the Perl 5.8.1 that got included with 10.3
> is a threaded Perl, and a version of threads that is neither backward
> compatible with 5.8.0 (or any prior perl), nor forward compatible
> (it was changed
good idea, I just think for others it should auto install the update
even if they don't know about it...fink-core is for project admins or
should i be on that list?
On 14-Nov-03, at 5:28 AM, Benjamin Reed wrote:
Perhaps, I posted to fink-core about it and wanted the maintainer's
(max's) input
I like having them as one list, though an other wouldn't hurt me, just
mean an other mailbox for me to check :)
On 13-Nov-03, at 10:59 PM, Michael G Schwern wrote:
Would other folks find this split useful? If the answer is largely no,
then yes, I'll write my own filter.
--
I agree here and have started on this code, it's just i have two branch
I'm working on already, once they get merged i can start a new branch
ad add auto perl version code.
On 14-Nov-03, at 8:28 AM, Max Horn wrote:
Small side note: how about changing the code to automatically insert
such a dep
couldn't you just use gnu ??
Source: gnu
On 14-Nov-03, at 8:18 AM, Benjamin Reed wrote:
+Source: ftp://alpha.gnu.org/gnu/coreutils/%n-%v.tar.bz2
---
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand
but what happens at compile time?? can altivec code be compiled on and
non altivec machine??
On 14-Nov-03, at 7:48 AM, Max Horn wrote:
And properly written Altivec should always be written such that it
automatically switches to using non-altivec code if altivec is not
present. That's how I do
Am Freitag, 14.11.03 um 14:53 Uhr schrieb Darian Lanx:
-BEGIN PGP SIGNED MESSAGE-
Hash: RIPEMD160
David R. Morrison wrote:
I was looking at this last night, as Benjamin already posted our
conversation I can only add this:
The bug is definately fixed in coreutil 0.5.91
... which is an
Randal L. Schwartz wrote:
Apple won't even be able to upgrade in the future. I wonder who the
Apple bonehead was that decided "threads are better!". Ugh.
Was it an apple engineer that did this? Because Perl 5.8.1 final
out-of-the-box defaults to the same thing that's in Apple's perl. I
don't
> "Max" == Max Horn <[EMAIL PROTECTED]> writes:
Max> Small side note: how about changing the code to automatically insert
Max> such a dependency if the type is "Perl: x.y" ? This way it can't be
Max> forgotten. Would there be any cases where this automatic behavior is
Max> not desirable?
Not
Things seem to work perfectly now.
Thanks a lot Dave for this very clarifying msg.
On Nov 14, 2003, at 2:30 PM, David R. Morrison wrote:
...
The scheme was disrupted somewhat when Apple compiled perl 5.8.1 for
10.3
with a different architecture flag. So for "Type: perl 5.8.1" we now
need
/sw/
With all that in mind, I think I'm going to split the increasingly misnamed
get_perl_arch_dir() into perl_archname, perl_version and perl_binary.
Hell, it might even make sense to start a Fink::Perl class to hold all this.
On Fri, Nov 14, 2003 at 08:30:37AM -0500, David R. Morrison wrote:
> Ther
[...]
By the way, the versioned perl module packages are supposed to state
a build-time dependency on "perl5.x.y-core", which guarantees that the
appropriate binary is present to compile the package.
Small side note: how about changing the code to automatically insert
such a dependency if the type
> Is libqt-mt from qt3-shlibs still linked with the xfree-86-4.2.1
> libfreetype? This would still break it for Panther users and force them
> to recompile qt3 themselves.
I rebuilt 0.6.2 from scratch, using the identical source files as in
0.6.1 (except for updates to fink, fink-prebinding, apt
Am Freitag, 14.11.03 um 06:39 Uhr schrieb TheSin:
acctually so the debs stay the same, fink needs to have a virtual pkg
for altivec and then have a lame-altivec version which would depend on
altivec and would could then use -faltivec...
I disagree. First off, -faltivec does *not* automatically m
On vendredi, nov 14, 2003, at 14:38 Europe/Paris, David R. Morrison
wrote:
I'm sorry to say that 0.6.2 binaries are, in almost every case,
identical
to 0.6.1 with one important exception: the files are now owned by root
rather than by UID 2011. (The incorrect ownership arose from a
"feature"
o
-BEGIN PGP SIGNED MESSAGE-
Hash: RIPEMD160
David R. Morrison wrote:
I was looking at this last night, as Benjamin already posted our
conversation I can only add this:
The bug is definately fixed in coreutil 0.5.91
Now the Apple supplied /bin/rm or /bin/mv do call the faulty
implemen
Dear Justin and Jean-Francois,
I updated the perl-module handling code in CVS last night. I hope that this
fixes all remaining problems with UpdatePOD. (Actually, it should be fixed
even without the changes Justin made, which I might revert at some point
since they complicate things a bit. But
> There is a theory now that this comes from the Fink fileutils package,
> although I don't see how it can, since you were using /bin/mv.
To echo this concern: I experienced the "mv" bug last night on a package.
I removed fileutils and rebuilt the package. It still failed due to the
"mv" bug, eve
Martin,
I'm sorry to say that 0.6.2 binaries are, in almost every case, identical
to 0.6.1 with one important exception: the files are now owned by root
rather than by UID 2011. (The incorrect ownership arose from a "feature"
of hfs+ which was unknown to me: it turns out that a non-root hfs+ volu
Michael G Schwern <[EMAIL PROTECTED]> wrote:
> On Thu, Nov 13, 2003 at 08:21:06PM -0800, David R. Morrison wrote:
> > Log Message:
> > Revise get_perl_dir_arch() and the places where it is used, correcting some
> > errors in UpdatePOD in case of Type: perl, and guaranteeing that the correct
> >
Martin Costabel wrote:
> I really want to believe it, but I still don't see the mechanism by
> which it can have this effect for some people and not for others (with
> apparently exactly the same system).
It is /bin/mv that is the problem, a user on irc reported that
installing fink's fileutils pac
Martin Costabel wrote:
But do you also have a reproducible case for mv? I don't find one.
Actually, I can't even reproduce the rm case (OSX 10.3.1)
costabel% tar zxf X11ForMacOSXSource-wwdc.tar.gz
costabel% /bin/rm -rf X11ForMacOSXSource-wwdc
costabel%
as in that irc conversation, panther's built-
Benjamin Reed wrote:
[]
The test case that I could reproduce 100% with was:
tar -xjf /sw/src/X430.tar.bz2
rm -rf xc
But do you also have a reproducible case for mv? I don't find one.
Actually, I can't even reproduce the rm case (OSX 10.3.1)
costabel% tar zxf X11ForMacOSXSource-wwdc.tar.gz
costabel
Martin Costabel wrote:
But the trailing slash problem must have a different cause. I am still
unable to reproduce it, although I tried with all kinds of Fink versions
of fileutils. It must be something in the environment. But I don't know
where to look.
The test case that I could reproduce 100%
TheSin wrote:
shouldn't we also make a new bundle release of fileutils with depends on
coreutils to make sure ppl upgrade that aren't reading this :)
Perhaps, I posted to fink-core about it and wanted the maintainer's
(max's) input before I actually started messing with things like that. :)
--
James Gibbs wrote:
[]
Martin, I don't have fileutils installed. I don't know what the
coreutils package is, nor do I find it anywhere.
coreutils is in 10.3/unstable now, replaces fileutils, textutils, and
shellutils. You don't really need it, it just replaces a bunch of OSX
Unix commands by thei
On Nov 14, 2003, at 3:15 AM, Martin Costabel wrote:
James Gibbs wrote:
[]
jgibbs% mkdir /dir1
jgibbs% mkdir /dir2
jgibbs% /bin/mv /dir1/ /dir2/
mv: rename /dir1/ to /dir2/dir1/: No such file or directory
jgibbs% which mv
/bin/mv
There is a theory now that this comes from the Fink fileutils package
James Gibbs wrote:
[]
jgibbs% mkdir /dir1
jgibbs% mkdir /dir2
jgibbs% /bin/mv /dir1/ /dir2/
mv: rename /dir1/ to /dir2/dir1/: No such file or directory
jgibbs% which mv
/bin/mv
There is a theory now that this comes from the Fink fileutils package,
although I don't see how it can, since you were us
Fred wrote:
Hi all,
I just wanted to report a bug, installing the vim-nox binary from a
fresh Fink 0.6.1 installs dlcompat as a dependency, but does not install
dlcompat-shlibs rendering it non-functional. Manually installing
dlcomat-shlibs gets everything going again.
I already suggested some
On Thu, 13 Nov 2003, Michael G Schwern wrote:
> On Fri, Nov 14, 2003 at 12:18:39AM -0500, Daniel Macks wrote:
> > On Wed, Nov 12, 2003 at 07:32:44PM -0800, Michael G Schwern wrote:
> > > I have a request. The amount of mail on fink-commits generated by .info
> > > and .patch file changes is rathe
47 matches
Mail list logo
