It's time for another 7.6 distribution soon. There are a couple of nasty little bugs that have been fixed since the last one, and I said we were going to release a bit more often. No big rush this time though, so if someone has something in perticular they'd like to fix it can wait a week. If not I'll try to make a release this weekend. Take a look at CHANGES and make sure your changes are documented.
On Fri, 2 Dec 2005 10:45:06 +0000 (UTC) "Peter Bortas @ Pike developers forum" 10353@lyskom.lysator.liu.se wrote:
It's time for another 7.6 distribution soon. There are a couple of nasty little bugs that have been fixed since the last one, and I said we were going to release a bit more often. No big rush this time though, so if someone has something in perticular they'd like to fix it can wait a week. If not I'll try to make a release this weekend. Take a look at CHANGES and make sure your changes are documented.
Looking through all the config.log files for OpenBSD specific problems that might need fixed, I found this:
configure:28435: checking sys/user.h usability configure:28447: /usr/ports/lang/pike/w-pike-7.6.50/build-i386/ smartlink cc -c -O2 -pipe -O2 -I/usr /local/include -I/usr/X11R6/ include -I/usr/ports/lang/pike/w-pike-7.6.50/Pike-v7.6.50/src -I/usr/ ports/lang/pike/w-pike-7.6.50/build-i386 conftest.c >&5 In file included from /usr/include/sys/user.h:41, from conftest.c:86: /usr/include/sys/ucred.h:46: error: `NGROUPS' undeclared here (not in a function) In file included from /usr/include/sys/sysctl.h:47, from /usr/include/sys/user.h:46, from conftest.c:86: /usr/include/sys/proc.h:61: error: `MAXLOGNAME' undeclared here (not in a function) /usr/include/sys/proc.h:217: error: `MAXCOMLEN' undeclared here (not in a function) In file included from /usr/include/sys/user.h:46, from conftest.c:86: /usr/include/sys/sysctl.h:328: error: `MAXLOGNAME' undeclared here (not in a function)
On OpenBSD, you have to include <sys/param.h> before you include <sys/user.h>.
Adam
7.7 builds have been broken since build 1217 and I can't backport this until 7.7 runs again. I suspect Nilssons export list checkin broke export.
On Fri, 2 Dec 2005 10:45:06 +0000 (UTC) "Peter Bortas @ Pike developers forum" 10353@lyskom.lysator.liu.se wrote:
It's time for another 7.6 distribution soon. There are a couple of nasty little bugs that have been fixed since the last one, and I said we were going to release a bit more often. No big rush this time though, so if someone has something in perticular they'd like to fix it can wait a week. If not I'll try to make a release this weekend. Take a look at CHANGES and make sure your changes are documented.
Also, the pdflib configure fix didn't quite fix the problem.
AC_CHECK_LIB(z, deflate, [ LIBS="${LIBS-} -lz" ]) AC_CHECK_LIB(jpeg,jpeg_CreateCompress, [ LIBS=${LIBS-} -ljpeg ]) AC_CHECK_LIB(png, png_read_image, [ LIBS="${LIBS-} -lpng" ]) AC_CHECK_LIB(tiff,TIFFOpen, [ LIBS="${LIBS-} -ltiff" ])
The jpeg test is missing its quotes (should be LIBS="${LIBS-} -ljpeg"). This is in both 7.6 and 7.7.
Adam
In the last episode (Dec 02), Adam Montague said:
On Fri, 2 Dec 2005 10:45:06 +0000 (UTC) "Peter Bortas @ Pike developers forum" 10353@lyskom.lysator.liu.se wrote:
It's time for another 7.6 distribution soon. There are a couple of nasty little bugs that have been fixed since the last one, and I said we were going to release a bit more often. No big rush this time though, so if someone has something in perticular they'd like to fix it can wait a week. If not I'll try to make a release this weekend. Take a look at CHANGES and make sure your changes are documented.
Also, the pdflib configure fix didn't quite fix the problem.
AC_CHECK_LIB(z, deflate, [ LIBS="${LIBS-} -lz" ]) AC_CHECK_LIB(jpeg,jpeg_CreateCompress, [ LIBS=${LIBS-} -ljpeg ]) AC_CHECK_LIB(png, png_read_image, [ LIBS="${LIBS-} -lpng" ]) AC_CHECK_LIB(tiff,TIFFOpen, [ LIBS="${LIBS-} -ltiff" ])
Out of curiosity, what's the purpose of the dash in ${LIBS-} ? If LIBS is unset, then $LIBS will expand to the empty string anyway.
Nilsson checked this in last friday. Besides the OpenBSD fix, is anything else missing?
The fink patches are not fully in.
I just figured something i may have not made clear about these patches.
While it's really nice to include changes that can made it in the mainstream distribution, the patch file provided mainly contains fink- specific changes. I didn't expected the patch to be applied to the mainstream source.
However, the pike7.6.patch file should be available alongside the pike7.6.info file for being able to build the package.
And the build is now utterly broken because of this.
FOO ?= bar
is _not_ portable Makeish.
On Fri, 2 Dec 2005 10:45:06 +0000 (UTC) "Peter Bortas @ Pike developers forum" 10353@lyskom.lysator.liu.se wrote:
It's time for another 7.6 distribution soon. There are a couple of nasty little bugs that have been fixed since the last one, and I said we were going to release a bit more often. No big rush this time though, so if someone has something in perticular they'd like to fix it can wait a week. If not I'll try to make a release this weekend. Take a look at CHANGES and make sure your changes are documented.
Here's another configure test that fails on OpenBSD:
configure:37568: checking if signal handlers reset automatically configure:37646: /home/adam/Pike-v7.6-snapshot/build/smartlink gcc -o conftest -g -O2 -pipe -Wa,--execstack -pthread -I/home/adam/Pike-v7.6- snapshot/build/bundles/include -I/usr/local/include -I/usr/X11R6/include -I/home/adam/Pike-v7.6-snapshot/src -I/home/adam/Pike-v7.6- snapshot/build -L/home/adam/Pike-v7.6-snapshot/build/bundles/lib -R/usr/local/lib -L/usr/local/lib -R/usr/X11R6/lib -L/usr/X11R6/lib -pthread conftest.c -lm >&5 configure: In function `main': configure:37627: error: storage size of `sentinel' isn't known configure:37649: $? = 1 configure: program exited with status 1
Its because #ifdef HAVE_POLL is true, but there is no #ifdef HAVE_POLL_H test to include poll.h, and poll.h is needed on OpenBSD.
Adam
Release candidate: http://pike.ida.liu.se/pub/pike/beta/7.6.56/Pike-v7.6.56.tar.gz
Hm, is it intentional that
A) The bundles are included _both_ as tar.gz and extracted
B) There are 5 versions of GMP bundled, two of which are extracted
?
There's also an extracted nettle-1.10 in bundles/. We probably need to fix export.pike so that it doesn't include directories under bundles.
And we should include nettle-1.13, not 1.10. The 1.10 bundle doesn't build on by Solaris box, but the 1.13 one does.
Not really, no. I did the release from a working copy this time and didn't know the bundles directory was included in full. I'll make another releases shortly and include grubbas autoconf fixes too.
I guess I should tell people that I did a release too... : http://pike.ida.liu.se/pub/pike/beta/7.6.58/Pike-v7.6.58.tar.gz
I have updated the xenofarm bundles to use nettle 1.14. Highlights:
- Faster MD5 on x86. - Less broken on sparc64. - Faster AES for everyone. - Faster RC4 on sparc.
I'm not sure that any of the computers in the xenofarm actually uses bundles, so it should be tested on a nettle-less computer. Obviously none of the 1.13 patches are applied.
I guess I should tell people that I did a release too... : http://pike.ida.liu.se/pub/pike/beta/7.6.58/Pike-v7.6.58.tar.gz
For those who would like to test this realease using fink, please use both included files.
More informations at <url:http://bertrand.gotpike.org/space/start/ 2005-12-29/1>
There's a minor typo in the man page introduced since the last distribution.
----8<----8<----8<----8<---- --- pike.1.ori 2005-12-29 11:35:33.000000000 +0100 +++ pike.1 2005-12-29 11:35:52.000000000 +0100 @@ -301,7 +301,7 @@ .B *(bd/tools/pike.el emacs fontlock definitions for pike-mode .TP -.B *(bd/include/pike/pike\n(mn.\n(mn/* +.B *(bd/include/pike/pike\n(mj.\n(mn/* C-include files .TP .B *(bd/man/* ---->8---->8---->8---->8----
On Mon, 19 Dec 2005 04:40:06 +0000 (UTC) "Peter Bortas @ Pike developers forum" 10353@lyskom.lysator.liu.se wrote:
Release candidate: http://pike.ida.liu.se/pub/pike/beta/7.6.56/Pike-v7.6.56.tar.gz
On OpenBSD, in modules/files I get:
configure: WARNING: sys/mount.h: present but cannot be compiled configure: WARNING: sys/mount.h: check for missing prerequisite headers?
Its because sys/mount.h needs sys/param.h included first.
Adam
On Mon, 19 Dec 2005 15:04:56 -0500 Adam Montague amontague@siriushosting.com wrote:
On Mon, 19 Dec 2005 04:40:06 +0000 (UTC) "Peter Bortas @ Pike developers forum" 10353@lyskom.lysator.liu.se wrote:
Release candidate: http://pike.ida.liu.se/pub/pike/beta/7.6.56/Pike-v7.6.56.tar.gz
On OpenBSD, in modules/files I get:
configure: WARNING: sys/mount.h: present but cannot be compiled configure: WARNING: sys/mount.h: check for missing prerequisite headers?
Its because sys/mount.h needs sys/param.h included first.
Adam
Also, in modules/Yp the tests for rpc/clnt.h and rpcsvc/yp_prot.h need to have rpc/rpc.h included first.
Adam
pike-devel@lists.lysator.liu.se
-
Adam Montague -
Bertrand LUPART -
Dan Nelson -
Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum -
Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum
-
Martin Nilsson (Opera) ＠ Pike (-) developers forum
-
Peter Bortas ＠ Pike developers forum