p5pRT
commented
13 years ago From alanh@fairlite.co.uk
Created by alanh@fairlite.co.uk
This patch adds support for the Atari FreeMiNT platform.
This is directly related to ticket 73786.
Closed p5pRT closed 10 years ago
p5pRT
commented
13 years ago This patch adds support for the Atari FreeMiNT platform.
This is directly related to ticket 73786.
p5pRT
commented
13 years ago Here's the patch.
Thanks\,
Alan.
p5pRT
commented
13 years ago p5pRT
commented
13 years ago On Thu\, 28 Apr 2011 20:37:05 +0100\, Alan Hourihane \alanh@​fairlite\.co\.uk wrote:
Here's the patch.
Sorry to say it is just too late for 5.14.0
Why exactly is this needed? As "" is the default. What other test would trigger it to be something else\, and in that case\, could that other test be amended by adding criteria?
@@ -1402\,7 +1405\,7 @@ xcat=/bin/cat test -f $xcat$_exe || xcat=/usr/bin/cat if test ! -f $xcat$_exe; then - for p in `echo $PATH | sed -e "s/$p_/ /g"` $paths; do + for p in `echo $PATH | sed -e 's/$p_/ /g'` $paths; do if test -f $p/cat$_exe; then xcat=$p/cat break
Looks sane\, but again not in time for 5.14.0. This is btw an unchanged item from the original "dist" metaconfig package. I'll contact Rafaël (the maintainer of dist) for his opinion\, as that unit is still unchanged in his svn repo. (git++ for giving me git svn clone)
Thanks\, Alan.
-- H.Merijn Brand http://tux.nl Perl Monger http://amsterdam.pm.org/ using 5.00307 through 5.12 and porting perl5.13.x on HP-UX 10.20\, 11.00\, 11.11\, 11.23 and 11.31\, OpenSuSE 10.1\, 11.0 .. 11.3 and AIX 5.2 and 5.3. http://mirrors.develooper.com/hpux/ http://www.test-smoke.org/ http://qa.perl.org http://www.goldmark.org/jeff/stupid-disclaimers/
p5pRT
commented
13 years ago The RT System itself - Status changed from 'new' to 'open'
p5pRT
commented
13 years ago Grrr! RT’s pop-up menus made me delete the freemint and 5.12.3 values.
p5pRT
commented
13 years ago Is there any chance this could be integrated before 5.16?
On Fri Apr 29 03:19:00 2011\, hmbrand wrote:
On Thu\, 28 Apr 2011 20:37:05 +0100\, Alan Hourihane \alanh@​fairlite\.co\.uk wrote:
Here's the patch.
Sorry to say it is just too late for 5.14.0
--- perl-5.12.3/Configure 2011-02-21 09:18:09.000000000 +0000 +++ perl-5.12.3-mint/Configure 2011-02-21 09:38:49.000000000 +0000 @@ -1275\,6 +1275\,9 @@ elif test -n "$DJGPP"; then : DOS DJGPP _exe=".exe" +elif test -f "/kern/cookiejar"; then +: MiNT + _exe="" elif test -d c:/. -o -n "$is_os2" ; then : OS/2 or cygwin _exe=".exe"
Why exactly is this needed? As "" is the default. What other test would trigger it to be something else\, and in that case\, could that other test be amended by adding criteria?
@@ -1402\,7 +1405\,7 @@ xcat=/bin/cat test -f $xcat$_exe || xcat=/usr/bin/cat if test ! -f $xcat$_exe; then - for p in `echo $PATH | sed -e "s/$p_/ /g"` $paths; do + for p in `echo $PATH | sed -e 's/$p_/ /g'` $paths; do if test -f $p/cat$_exe; then xcat=$p/cat break
Looks sane\, but again not in time for 5.14.0. This is btw an unchanged item from the original "dist" metaconfig package. I'll contact Rafa�l (the maintainer of dist) for his opinion\, as that unit is still unchanged in his svn repo. (git++ for giving me git svn clone)
Thanks\, Alan.
p5pRT
commented
13 years ago Is there any chance this could be integrated before 5.16?
On Fri Apr 29 03:19:00 2011\, hmbrand wrote:
On Thu\, 28 Apr 2011 20:37:05 +0100\, Alan Hourihane \alanh@​fairlite\.co\.uk wrote:
Here's the patch.
Sorry to say it is just too late for 5.14.0
--- perl-5.12.3/Configure 2011-02-21 09:18:09.000000000 +0000 +++ perl-5.12.3-mint/Configure 2011-02-21 09:38:49.000000000 +0000 @@ -1275\,6 +1275\,9 @@ elif test -n "$DJGPP"; then : DOS DJGPP _exe=".exe" +elif test -f "/kern/cookiejar"; then +: MiNT + _exe="" elif test -d c:/. -o -n "$is_os2" ; then : OS/2 or cygwin _exe=".exe"
Why exactly is this needed? As "" is the default. What other test would trigger it to be something else\, and in that case\, could that other test be amended by adding criteria?
@@ -1402\,7 +1405\,7 @@ xcat=/bin/cat test -f $xcat$_exe || xcat=/usr/bin/cat if test ! -f $xcat$_exe; then - for p in `echo $PATH | sed -e "s/$p_/ /g"` $paths; do + for p in `echo $PATH | sed -e 's/$p_/ /g'` $paths; do if test -f $p/cat$_exe; then xcat=$p/cat break
Looks sane\, but again not in time for 5.14.0. This is btw an unchanged item from the original "dist" metaconfig package. I'll contact Rafa�l (the maintainer of dist) for his opinion\, as that unit is still unchanged in his svn repo. (git++ for giving me git svn clone)
Thanks\, Alan.
p5pRT
commented
13 years ago On Sun\, 16 Oct 2011 17:41:18 -0700\, "Father Chrysostomos via RT" \perlbug\-comment@​perl\.org wrote:
Is there any chance this could be integrated before 5.16?
I don't think so\, as there was no reply at all on the first part\, and the second part is plain wrong. I tested it after I posted the "sane" comment.
On Fri Apr 29 03:19:00 2011\, hmbrand wrote:
On Thu\, 28 Apr 2011 20:37:05 +0100\, Alan Hourihane \alanh@​fairlite\.co\.uk wrote:
Here's the patch.
Sorry to say it is just too late for 5.14.0
--- perl-5.12.3/Configure 2011-02-21 09:18:09.000000000 +0000 +++ perl-5.12.3-mint/Configure 2011-02-21 09:38:49.000000000 +0000 @@ -1275\,6 +1275\,9 @@ elif test -n "$DJGPP"; then : DOS DJGPP _exe=".exe" +elif test -f "/kern/cookiejar"; then +: MiNT + _exe="" elif test -d c:/. -o -n "$is_os2" ; then : OS/2 or cygwin _exe=".exe"
Why exactly is this needed? As "" is the default. What other test would trigger it to be something else\, and in that case\, could that other test be amended by adding criteria?
@@ -1402\,7 +1405\,7 @@ xcat=/bin/cat test -f $xcat$_exe || xcat=/usr/bin/cat if test ! -f $xcat$_exe; then - for p in `echo $PATH | sed -e "s/$p_/ /g"` $paths; do + for p in `echo $PATH | sed -e 's/$p_/ /g'` $paths; do if test -f $p/cat$_exe; then xcat=$p/cat break
Looks sane\, but again not in time for 5.14.0. This is btw an unchanged item from the original "dist" metaconfig package. I'll contact Rafa�l (the maintainer of dist) for his opinion\, as that unit is still unchanged in his svn repo. (git++ for giving me git svn clone)
Thanks\, Alan.
-- H.Merijn Brand http://tux.nl Perl Monger http://amsterdam.pm.org/ using 5.00307 through 5.14 and porting perl5.15.x on HP-UX 10.20\, 11.00\, 11.11\, 11.23 and 11.31\, OpenSuSE 10.1\, 11.0 .. 11.4 and AIX 5.2 and 5.3. http://mirrors.develooper.com/hpux/ http://www.test-smoke.org/ http://qa.perl.org http://www.goldmark.org/jeff/stupid-disclaimers/
p5pRT
commented
11 years ago On Sun Oct 16 22:56:47 2011\, hmbrand wrote:
On Sun\, 16 Oct 2011 17:41:18 -0700\, "Father Chrysostomos via RT" \perlbug\-comment@​perl\.org wrote:
Is there any chance this could be integrated before 5.16?
I don't think so\, as there was no reply at all on the first part\, and the second part is plain wrong. I tested it after I posted the "sane" comment.
There's a followup for the first part at:
http://www.nntp.perl.org/group/perl.perl5.porters/2012/08/msg191018.html
which doesn't seem to have been picked up by RT.
Presumably the cygwin test for -d c:/. is matching.
Tony
p5pRT
commented
11 years ago What is the logic behind the second part of the patch? It only appears to be replacing double quotes with single quotes. On any Unix-like system\, this is clearly wrong\, since single quotes would prevent the $p from being interpolated into the string. Does FreeMiNT have different semantics?
Deven
On Mon\, Jun 24\, 2013 at 1:35 AM\, Tony Cook via RT \<perlbug-followup@perl.org
wrote:
On Sun Oct 16 22:56:47 2011\, hmbrand wrote:
On Sun\, 16 Oct 2011 17:41:18 -0700\, "Father Chrysostomos via RT" \perlbug\-comment@​perl\.org wrote:
Is there any chance this could be integrated before 5.16?
I don't think so\, as there was no reply at all on the first part\, and the second part is plain wrong. I tested it after I posted the "sane" comment.
There's a followup for the first part at:
http://www.nntp.perl.org/group/perl.perl5.porters/2012/08/msg191018.html
which doesn't seem to have been picked up by RT.
Presumably the cygwin test for -d c:/. is matching.
Tony
--- via perlbug: queue: perl5 status: open https://rt-archive.perl.org/perl5/Ticket/Display.html?id=89502
p5pRT
commented
11 years ago On Thu\, Jul 4\, 2013 at 6:36 AM\, Alan Hourihane \alanh@​fairlite\.co\.uk wrote:
No. FreeMiNT doesn't have different semantics.
Maybe it's a bash 4.2 thing. I'd need to check.
The distinction between single and double quotes goes back decades to the original Bourne shell. Every version of bash will treat "$" inside single quotes as literal text.
What was broken about the Configure script that the patch was attempting to fix?
Deven
p5pRT
commented
11 years ago On Thu Apr 28 12:32:18 2011\, alanh wrote:
This patch adds support for the Atari FreeMiNT platform.
This is directly related to ticket 73786.
See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2013-09/msg00174.html relating to dl_dld.xs
Nicholas Clark
p5pRT
commented
11 years ago On Fri Apr 29 03:19:00 2011\, hmbrand wrote:
On Thu\, 28 Apr 2011 20:37:05 +0100\, Alan Hourihane \alanh@​fairlite\.co\.uk wrote:
Here's the patch.
Sorry to say it is just too late for 5.14.0
--- perl-5.12.3/Configure 2011-02-21 09:18:09.000000000 +0000 +++ perl-5.12.3-mint/Configure 2011-02-21 09:38:49.000000000 +0000 @@ -1275\,6 +1275\,9 @@ elif test -n "$DJGPP"; then : DOS DJGPP _exe=".exe" +elif test -f "/kern/cookiejar"; then +: MiNT + _exe="" elif test -d c:/. -o -n "$is_os2" ; then : OS/2 or cygwin _exe=".exe"
Why exactly is this needed? As "" is the default. What other test would trigger it to be something else\, and in that case\, could that other test be amended by adding criteria?
@@ -1402\,7 +1405\,7 @@ xcat=/bin/cat test -f $xcat$_exe || xcat=/usr/bin/cat if test ! -f $xcat$_exe; then - for p in `echo $PATH | sed -e "s/$p_/ /g"` $paths; do + for p in `echo $PATH | sed -e 's/$p_/ /g'` $paths; do if test -f $p/cat$_exe; then xcat=$p/cat break
Looks sane\, but again not in time for 5.14.0. This is btw an unchanged item from the original "dist" metaconfig package. I'll contact Rafa�l (the maintainer of dist) for his opinion\, as that unit is still unchanged in his svn repo. (git++ for giving me git svn clone)
Thanks\, Alan.
Just to go over the Configure patch again.
The first hunk is required because otherwise the OS2/Cygwin test succeeds on this platform\, and that's incorrect.
The second hunk isn't required. I've just retested without it.
Can this be committed without that hunk ?
Also\, we need the above comment rectifying so we don't lose dld support.
Thanks.
p5pRT
commented
11 years ago On Mon\, 16 Sep 2013 14:16:02 -0700\, "Alan Hourihane via RT" \perlbug\-followup@​perl\.org wrote:
On Fri Apr 29 03:19:00 2011\, hmbrand wrote:
On Thu\, 28 Apr 2011 20:37:05 +0100\, Alan Hourihane \alanh@​fairlite\.co\.uk wrote:
Here's the patch.
Sorry to say it is just too late for 5.14.0
--- perl-5.12.3/Configure 2011-02-21 09:18:09.000000000 +0000 +++ perl-5.12.3-mint/Configure 2011-02-21 09:38:49.000000000 +0000 @@ -1275\,6 +1275\,9 @@ elif test -n "$DJGPP"; then : DOS DJGPP _exe=".exe" +elif test -f "/kern/cookiejar"; then ^ ^
As part of the original patch is about quotes\, I might add that those are not needed here\, right?
+: MiNT + _exe="" elif test -d c:/. -o -n "$is_os2" ; then : OS/2 or cygwin _exe=".exe"
Why exactly is this needed? As "" is the default. What other test would trigger it to be something else\, and in that case\, could that other test be amended by adding criteria?
@@ -1402\,7 +1405\,7 @@ xcat=/bin/cat test -f $xcat$_exe || xcat=/usr/bin/cat if test ! -f $xcat$_exe; then - for p in `echo $PATH | sed -e "s/$p_/ /g"` $paths; do + for p in `echo $PATH | sed -e 's/$p_/ /g'` $paths; do if test -f $p/cat$_exe; then xcat=$p/cat break
Looks sane\, but again not in time for 5.14.0. This is btw an unchanged item from the original "dist" metaconfig package. I'll contact Rafa�l (the maintainer of dist) for his opinion\, as that unit is still unchanged in his svn repo. (git++ for giving me git svn clone)
Thanks\, Alan.
Just to go over the Configure patch again.
The first hunk is required because otherwise the OS2/Cygwin test succeeds on this platform\, and that's incorrect.
The second hunk isn't required. I've just retested without it.
Then do not change it. This is most likely what was holding the patch from being applied anyway. Here is the comments (slightly simplified) from the maintainer of meta\, just for some clarification: --8\<--- The correct line is of course the one that reads:
for p in `echo $PATH | sed -e "s/$p_/ /g"` $paths; do
but this patch seems to prefer the one where the substitution pattern would be enclosed in simple quotes. That is of course wrong\, since the then $p_ variable will not be expanded (p_ is the path separator\, as computed in Head.U).
In the SVN repository\, the line present in spitshell.U is the correct one. -->8---
Can this be committed without that hunk ?
I think so
Also\, we need the above comment rectifying so we don't lose dld support.
Thanks.
-- H.Merijn Brand http://tux.nl Perl Monger http://amsterdam.pm.org/ using perl5.00307 .. 5.19 porting perl5 on HP-UX\, AIX\, and openSUSE http://mirrors.develooper.com/hpux/ http://www.test-smoke.org/ http://qa.perl.org http://www.goldmark.org/jeff/stupid-disclaimers/
p5pRT
commented
11 years ago On Mon\, Sep 16\, 2013 at 02:16:02PM -0700\, Alan Hourihane via RT wrote:
Also\, we need the above comment rectifying so we don't lose dld support.
The dl_dld.xs file makes no use of any of the probing done by Configure - for example\, it unconditionally includes \<dld.h> instead of wrapping it in C pre-processor macros. So we don't need to re-instate the probes.
Also\, no other platform is using dld\, so if we put the file back as dl_mint.xs\, I believe that Configure (as-is) will automatically pick it up for MiNT. ($osname is set to 'mint'\, correct?)
That keeps Configure simpler\, without stopping MiNT working.
Nicholas Clark
p5pRT
commented
11 years ago On 09/17/13 08:12\, Nicholas Clark via RT wrote:
On Mon\, Sep 16\, 2013 at 02:16:02PM -0700\, Alan Hourihane via RT wrote:
Also\, we need the above comment rectifying so we don't lose dld support. The dl_dld.xs file makes no use of any of the probing done by Configure - for example\, it unconditionally includes \<dld.h> instead of wrapping it in C pre-processor macros. So we don't need to re-instate the probes.
Also\, no other platform is using dld\, so if we put the file back as dl_mint.xs\, I believe that Configure (as-is) will automatically pick it up for MiNT. ($osname is set to 'mint'\, correct?)
That keeps Configure simpler\, without stopping MiNT working.
Great. Thanks.
Let me know when to grab something to test.
Alan.
p5pRT
commented
11 years ago On Tue Sep 17 01:24:49 2013\, alanh wrote:
Let me know when to grab something to test.
Alan.
To move this ticket along.
As a 3rd party who understands very little of this ticket\, I think you need to remove the double to single quotes change in Configure. Also your patch is a diff patch\, not a git patch. Here at p5p\, we prefer\, but don't require\, git patches with commit messages. Mention this ticket number in the commit message body. Also since dld was removed per http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2013-09/msg00174.html \, this patch seems uncommitable now. Note the OP diff patch is from 5.12\, not 5.19.
Also\, some [of my] questions to answer either here in this ticket or in commit message.
Why did the fixed stack size change in the OS?
____________________________________________________
Why more error cases for dld?
-- bulk88 ~ bulk88 at hotmail.com
p5pRT
commented
11 years ago On 10/30/13 03:16\, bulk88 via RT wrote:
To move this ticket along.
As a 3rd party who understands very little of this ticket\, I think you need to remove the double to single quotes change in Configure. Also your patch is a diff patch\, not a git patch. Here at p5p\, we prefer\, but don't require\, git patches with commit messages. Mention this ticket number in the commit message body. Also since dld was removed per http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2013-09/msg00174.html \, this patch seems uncommitable now. Note the OP diff patch is from 5.12\, not 5.19.
As previous tickets have mentioned that dld will be added back with a freemint renamed file. And as mentioned the Configure quotes can be reverted.
Also\, some [of my] questions to answer either here in this ticket or in commit message.
Why did the fixed stack size change in the OS?
Because the app needs more stack.
____________________________________________________ diff -ur perl-5.12.3/ext/DynaLoader/dl_dld.xs perl-5.12.3-mint/ext/DynaLoader/dl_dld.xs --- perl-5.12.3/ext/DynaLoader/dl_dld.xs 2010-11-21 21:40:55.000000000 +0000 +++ perl-5.12.3-mint/ext/DynaLoader/dl_dld.xs 2011-02-21 09:38:51.000000000 +0000 @@ -116\,6 +116\,18 @@ goto haverror; }
+ DLDEBUG(1\,PerlIO_printf(Perl_debug_log\, "dld_link(libm.a)\n")); + if (dlderr = dld_link("/usr/lib/libm.a")) { + SaveError(aTHX_ "dld_link(libm.a): %s"\, dld_strerror(dlderr)); + goto haverror; + } + + DLDEBUG(1\,PerlIO_printf(Perl_debug_log\, "dld_link(libc.a)\n")); + if (dlderr = dld_link("/usr/lib/libc.a")) { + SaveError(aTHX_ "dld_link(libc.a): %s"\, dld_strerror(dlderr)); + goto haverror; + } + max = AvFILL(dl_resolve_using); for (x = 0; x \<= max; x++) { char *sym = SvPVX(*av_fetch(dl_resolve_using\, x\, 0)); ____________________________________________________ Why more error cases for dld?
Because dld needs to link with system libraries.
Alan.
p5pRT
commented
11 years ago On Tue Sep 17 01:24:49 2013\, alanh wrote:
Let me know when to grab something to test.
I've pushed an update to tonyc/freemint reinstating dld renamed as dld_freemint\, *and* with the extra dld_links for libc and libm from your original patch.
You can download a snapshot to test from:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
Tony
p5pRT
commented
11 years ago On 11/11/13 00:40\, Tony Cook via RT wrote:
On Tue Sep 17 01:24:49 2013\, alanh wrote:
Let me know when to grab something to test. I've pushed an update to tonyc/freemint reinstating dld renamed as dld_freemint\, *and* with the extra dld_links for libc and libm from your original patch.
You can download a snapshot to test from:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
Thanks Tony\, I'll grab it and report back.
Alan.
p5pRT
commented
11 years ago Hi Tony\,
So I'm building a copy now and get these warnings when building dl_freemint.xs...
../../miniperl "-I../../lib" "-I../../lib" ../../lib/ExtUtils/xsubpp -noprototypes -typemap ../../lib/ExtUtils/typemap DynaLoader.xs > DynaLoader.xsc && mv DynaLoader.xsc DynaLoader.c Warning: Found a 'CODE' section which seems to be using 'RETVAL' but no 'OUTPUT' section. in DynaLoader.xs\, line 146 Warning: Found a 'CODE' section which seems to be using 'RETVAL' but no 'OUTPUT' section. in DynaLoader.xs\, line 163 cc -c -D_GNU_SOURCE -fno-strict-aliasing -pipe -O -DVERSION=\"1.21\" -DXS_VERSION=\"1.21\" "-I../.." -DLIBC="/usr/lib/libc.a" DynaLoader.c DynaLoader.xs: In function 'dl_private_init': DynaLoader.xs:74:15: warning: initialization discards 'const' qualifier from pointer target type [enabled by default]
We're just missing "const" on line 74\, but I'm not sure how line 146 & 163 should be fixed ?
Also\, we can remove the __linux__ references in this file too.
I need to send a new files for FreeMiNT hints and another minor patch that I'll update next.
Alan.
p5pRT
commented
11 years ago Attached is a patch against the renamed dl_freemint.xs file.
p5pRT
commented
11 years ago p5pRT
commented
11 years ago Attached is a patch to ext/POSIX/Makefile.PL to stop including libm for freemint as well.
p5pRT
commented
11 years ago p5pRT
commented
11 years ago Finally a hints file for FreeMiNT.
p5pRT
commented
11 years ago p5pRT
commented
11 years ago Actually\, scrap that last freemint.sh hints file and use this as I've updated the comments.
p5pRT
commented
11 years ago p5pRT
commented
11 years ago On Thu Nov 14 08:51:52 2013\, alanh wrote:
Actually\, scrap that last freemint.sh hints file and use this as I've updated the comments.
I'm confused by this.
Your hints file enabled shared libraries through dl_freeminit.xs\, but sets "so" to 'none' - does that mean non-XS shared libraries are not supported?
Other comments:
1) Instead of setting malloc_cflags\, you might want to set usemymalloc=n\, so perl uses the system malloc instead of its own\, though possibly you could use both so someone could enable usemymalloc with -Dusemymalloc=y at Configure time:
case "$usemymalloc" in '') usemymalloc='n' ;; esac malloc_cflags='ccflags="-DUSE_PERL_SBRK -DPERL_SBRK_VIA_MALLOC $ccflags"'
2) are you writing a README.mint file? The hints file refers to one\, but blead doesn't include it.
Tony
p5pRT
commented
11 years ago On Thu Nov 14 08:51:52 2013\, alanh wrote:
Actually\, scrap that last freemint.sh hints file and use this as I've updated the comments.
I've pushed an updated branch with your changes.
The same link:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
should download the new changes.
Tony
p5pRT
commented
11 years ago On 11/15/13 00:47\, Tony Cook via RT wrote:
On Thu Nov 14 08:51:52 2013\, alanh wrote:
Actually\, scrap that last freemint.sh hints file and use this as I've updated the comments. I'm confused by this.
Your hints file enabled shared libraries through dl_freeminit.xs\, but sets "so" to 'none' - does that mean non-XS shared libraries are not supported?
That's correct. The system doesn't have generalized shared libraries. That's why we only support XS shared objects through our dld interface.
Other comments:
1) Instead of setting malloc_cflags\, you might want to set usemymalloc=n\, so perl uses the system malloc instead of its own\, though possibly you could use both so someone could enable usemymalloc with -Dusemymalloc=y at Configure time:
case "$usemymalloc" in '') usemymalloc='n' ;; esac malloc_cflags='ccflags="-DUSE_PERL_SBRK -DPERL_SBRK_VIA_MALLOC $ccflags"'
Gotcha. I see darwin does this too\, so I'll add the case/esac clause.
2) are you writing a README.mint file? The hints file refers to one\, but blead doesn't include it.
No\, that reference can be dropped.
Alan.
p5pRT
commented
11 years ago On 11/15/13 01:01\, Tony Cook via RT wrote:
On Thu Nov 14 08:51:52 2013\, alanh wrote:
Actually\, scrap that last freemint.sh hints file and use this as I've updated the comments. I've pushed an updated branch with your changes.
The same link:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
should download the new changes.
Thanks\, I'm compiling now.
Alan.
p5pRT
commented
11 years ago Hi Tony\,
Final freemint.sh hints patch attached.
There's a few test failures\, but they're nothing to do with the build process now\, looks like they are libc/system level bugs.
Thanks\,
Alan.
p5pRT
commented
11 years ago p5pRT
commented
11 years ago On Fri Nov 15 05:52:47 2013\, alanh wrote:
Hi Tony\,
Final freemint.sh hints patch attached.
There's a few test failures\, but they're nothing to do with the build process now\, looks like they are libc/system level bugs.
I've applied that change\, squashing it into the original hints change\, and pushed back to the same branch\, please try:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
again.
Thanks\, Tony
p5pRT
commented
11 years ago On 11/17/13 23:04\, Tony Cook via RT wrote:
On Fri Nov 15 05:52:47 2013\, alanh wrote:
Hi Tony\,
Final freemint.sh hints patch attached.
There's a few test failures\, but they're nothing to do with the build process now\, looks like they are libc/system level bugs. I've applied that change\, squashing it into the original hints change\, and pushed back to the same branch\, please try:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
All good Tony. Thanks !
p5pRT
commented
11 years ago On Mon Nov 18 01:54:45 2013\, alanh wrote:
On 11/17/13 23:04\, Tony Cook via RT wrote:
On Fri Nov 15 05:52:47 2013\, alanh wrote:
Hi Tony\,
Final freemint.sh hints patch attached.
There's a few test failures\, but they're nothing to do with the build process now\, looks like they are libc/system level bugs. I've applied that change\, squashing it into the original hints change\, and pushed back to the same branch\, please try:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
All good Tony. Thanks !
Can you summarize how well FreeMINT support works for use in perldelta?
eg. Perl now builds on FreeMINT but fails (many|a few|specific) tests.
Tony
p5pRT
commented
11 years ago On Mon\, Nov 18\, 2013 at 8:38 PM\, Tony Cook via RT \<perlbug-followup@perl.org
wrote:
On Mon Nov 18 01:54:45 2013\, alanh wrote:
On 11/17/13 23:04\, Tony Cook via RT wrote:
On Fri Nov 15 05:52:47 2013\, alanh wrote:
Hi Tony\,
Final freemint.sh hints patch attached.
There's a few test failures\, but they're nothing to do with the build process now\, looks like they are libc/system level bugs. I've applied that change\, squashing it into the original hints change\, and pushed back to the same branch\, please try:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
All good Tony. Thanks !
Can you summarize how well FreeMINT support works for use in perldelta?
eg. Perl now builds on FreeMINT but fails (many|a few|specific) tests.
I went through those steps for Android a while back; if I recall correctly\, what more or else needs doing for a "new" platform is: An entry in Perl::OSType. An update to perlport.pod\, at least to the list of supported OSs\, and optionally to the main body of the file if there's any particular gotcha's in the system. An entry in perldelta. A README.platform\, particularly if it requires some particular building steps or some build types are not possible/problematic (e.g. threads\, shared libperl) Possibly an entry in Porting/todo.pod if some parts of the port need some love?
p5pRT
commented
10 years ago On Tue\, Sep 17\, 2013 at 9:11 AM\, Nicholas Clark \nick@​ccl4\.org wrote:
On Mon\, Sep 16\, 2013 at 02:16:02PM -0700\, Alan Hourihane via RT wrote:
Also\, we need the above comment rectifying so we don't lose dld support.
The dl_dld.xs file makes no use of any of the probing done by Configure - for example\, it unconditionally includes \<dld.h> instead of wrapping it in C pre-processor macros. So we don't need to re-instate the probes.
Also\, no other platform is using dld\, so if we put the file back as dl_mint.xs\, I believe that Configure (as-is) will automatically pick it up for MiNT. ($osname is set to 'mint'\, correct?)
That keeps Configure simpler\, without stopping MiNT working.
Nicholas Clark
ExtUtils::Mkbootstrap contains some specific code for «$Config{'dlsrc'} =~ /^dl_dld/». I have no idea if this should be updated or simply removed\, someone with more clue of dld/mint should take a look at it. It might explain the «libc/system level bugs» mentioned below.
Leon
p5pRT
commented
10 years ago On Mon Nov 18 15:38:39 2013\, tonyc wrote:
On Mon Nov 18 01:54:45 2013\, alanh wrote:
On 11/17/13 23:04\, Tony Cook via RT wrote:
On Fri Nov 15 05:52:47 2013\, alanh wrote:
Hi Tony\,
Final freemint.sh hints patch attached.
There's a few test failures\, but they're nothing to do with the build process now\, looks like they are libc/system level bugs. I've applied that change\, squashing it into the original hints change\, and pushed back to the same branch\, please try:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
All good Tony. Thanks !
Can you summarize how well FreeMINT support works for use in perldelta?
eg. Perl now builds on FreeMINT but fails (many|a few|specific) tests.
Hi Alan\,
Any chance of getting this summary?
Tony
p5pRT
commented
10 years ago On 01/15/14 03:54\, Tony Cook via RT wrote:
On Mon Nov 18 15:38:39 2013\, tonyc wrote:
On Mon Nov 18 01:54:45 2013\, alanh wrote:
On 11/17/13 23:04\, Tony Cook via RT wrote:
On Fri Nov 15 05:52:47 2013\, alanh wrote:
Hi Tony\,
Final freemint.sh hints patch attached.
There's a few test failures\, but they're nothing to do with the build process now\, looks like they are libc/system level bugs. I've applied that change\, squashing it into the original hints change\, and pushed back to the same branch\, please try:
http://perl5.git.perl.org/perl.git/snapshot/refs/heads/tonyc/freemint.tar.gz
All good Tony. Thanks ! Can you summarize how well FreeMINT support works for use in perldelta?
eg. Perl now builds on FreeMINT but fails (many|a few|specific) tests. Hi Alan\,
Any chance of getting this summary?
Yes\, sorry Tony.
"Perl now builds with FreeMiNT/Atari. It fails a few tests\, that needs some investigation. The FreeMiNT port uses GNU dld for loadable module capabilities. So ensure you have that library installed when building perl."
That's all really. Nothing more needed.
Alan.
p5pRT
commented
10 years ago On Wed Jan 15 04:11:29 2014\, alanh wrote:
"Perl now builds with FreeMiNT/Atari. It fails a few tests\, that needs some investigation. The FreeMiNT port uses GNU dld for loadable module capabilities. So ensure you have that library installed when building perl."
That's all really. Nothing more needed.
All the various patches applied as:
817bb5726ea19a5cf60d1aa249e44880d3267fab dld support for FreeMINT f48d8d9daf0f92cffbbef9a80e5f2c4d260c320c your changes cbc0b8140a1ef2e1a7119bb461f48395a4a8fc4d remove -lm -lposix fa1b2c4275e53adb349f1a988aec8a194d8763f3 hints file 16ef6ddfef6c2f8ca35702aceff4a7d27ce94c01 welcome to AUTHORS dcfa7505f02754d9e15a4fdc725df3b60cf57c06 a note in perlport.
Tony
p5pRT
commented
10 years ago On 01/16/14 04:41\, Tony Cook via RT wrote:
On Wed Jan 15 04:11:29 2014\, alanh wrote:
"Perl now builds with FreeMiNT/Atari. It fails a few tests\, that needs some investigation. The FreeMiNT port uses GNU dld for loadable module capabilities. So ensure you have that library installed when building perl."
That's all really. Nothing more needed. All the various patches applied as:
817bb5726ea19a5cf60d1aa249e44880d3267fab dld support for FreeMINT f48d8d9daf0f92cffbbef9a80e5f2c4d260c320c your changes cbc0b8140a1ef2e1a7119bb461f48395a4a8fc4d remove -lm -lposix fa1b2c4275e53adb349f1a988aec8a194d8763f3 hints file 16ef6ddfef6c2f8ca35702aceff4a7d27ce94c01 welcome to AUTHORS dcfa7505f02754d9e15a4fdc725df3b60cf57c06 a note in perlport.
Thanks Tony !
Alan.
p5pRT
commented
10 years ago @tonycoz - Status changed from 'open' to 'resolved'
Migrated from rt.perl.org#89502 (status was 'resolved')
Searchable as RT89502$