You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@apr.apache.org by "William A. Rowe, Jr." <wr...@rowe-clan.net> on 2008/08/09 16:37:14 UTC
1.3.3 tarballs for review
Available in the usual; http://apr.apache.org/dev/dist/, win32
with exported .mak files to follow as soon as I have a bit of
free time, busy weekend here so likely on Sunday eve (when I'll
look also at committing Carlo's efforts on MinGW).
+/-1
[ ] Release apr-1.3.3
[ ] Release apr-util-1.3.3
Re: 1.3.3 tarballs for review
Posted by Bojan Smojver <bo...@rexursive.com>.
On Sun, 2008-08-10 at 02:34 +0200, Rainer Jung wrote:
> Failed Tests Total Fail Failed %
> ===================================================
> testsock 8 1 12.50%
> Line 92: Problem getting http service (2): No such file or directory
>
> That's ignorable, http is not contained in /etc/services for Solaris 8.
We can trivially fix this by looking for something like telnet instead.
I'm guessing Solaris 8 has it in /etc/services, right?
--
Bojan
Re: 1.3.3 tarballs for review
Posted by Rainer Jung <ra...@kippdata.de>.
Tested on Solaris 8 Sparc, compiled with gcc 4.1.2.
Summary: only minor comments, so an non-binding
[+1] Release apr-1.3.3
[+1] Release apr-util-1.3.3
A) configure
============
1) Warnings
-----------
Two (old) warnings for apr-util
config.status: WARNING:
Makefile.in seems to ignore the --datarootdir setting
config.status: WARNING:
lib/Makefile.in seems to ignore the --datarootdir setting
Another similar warning for apr
config.status:1279: WARNING:
'apr-config.in' seems to ignore the --datarootdir setting
has been fixed between 1.3.0 and 1.3.3, most likely due to this new line:
diff apr-1.3.0/apr-config.in apr-1.3.3/apr-config.in
27a28
> datarootdir="@datadir@"
2) Missing AC_MSG_RESULT
------------------------
There is a new check AC_MSG_CHECKING(if fcntl returns EACCES when
F_SETLK is already held), which isn't combined with any AC_MSG_RESULT.
So the message
checking if fcntl returns EACCES when F_SETLK is already held...
has no result or newline at it's end and the output line directly
preoceeds with
checking if all interprocess locks affect threads... no
3) pid_t format in Solaris
--------------------------
Detection problem is fixed.
B) make
=======
Nothing special.
C) make test
============
1) apr
------
Failed Tests Total Fail Failed %
===================================================
testsock 8 1 12.50%
Line 92: Problem getting http service (2): No such file or directory
That's ignorable, http is not contained in /etc/services for Solaris 8.
testsockets 7 1 14.29%
Line 131: Could not bind socket (126): Cannot assign requested address
bind(12, 0x00126DB8, 32, 3) Err#126 EADDRNOTAVAIL
AF_INET6 name = ::1 port = 7772
This testsockets result is identical to version 1.3.0.
2) apr-util
-----------
Failed Tests Total Fail Failed %
===================================================
testxlate 1 1 100.00%
Line 63: expected <0>, but saw <22>
This result is identical to version 1.3.0.
Regards,
Rainer
Re: 1.3.3 tarballs for review
Posted by Mladen Turk <mt...@apache.org>.
William A. Rowe, Jr. wrote:
> Available in the usual; http://apr.apache.org/dev/dist/, win32
> with exported .mak files to follow as soon as I have a bit of
> free time, busy weekend here so likely on Sunday eve (when I'll
> look also at committing Carlo's efforts on MinGW).
>
> +/-1
> [+1] Release apr-1.3.3
> [+1] Release apr-util-1.3.3
>
Win32 (All tests passed.)
Well, except some warnings (with testall -v)
testpoll : Line 570: pollcb interface not supported
Compiles cleanly on WIN64/x86_64 and WIN64/IA64
Regards
--
^(TM)
Re: 1.3.3 tarballs for review
Posted by Ruediger Pluem <rp...@apache.org>.
On 08/09/2008 04:37 PM, William A. Rowe, Jr. wrote:
> Available in the usual; http://apr.apache.org/dev/dist/, win32
> with exported .mak files to follow as soon as I have a bit of
> free time, busy weekend here so likely on Sunday eve (when I'll
> look also at committing Carlo's efforts on MinGW).
>
> +/-1
> [ +1 ] Release apr-1.3.3
> [ +1 ] Release apr-util-1.3.3
>
>
Signatures and md5's good. Tested and passed all tests on
Linux (x86, 32 Bit) : RHEL3, RHEL4, OpenSuSE 10.2
Linux (AMD, 64 Bit) : RHEL4, RHEL5, OpenSuSE 10.1
Solaris (SPARC, 32 Bit): 8, 9, 10
Regards
RĂ¼diger
Re: 1.3.3 tarballs for review
Posted by Eric Covener <co...@gmail.com>.
Summary: no regression from 1.3.2 on platforms below. No apr-iconv in
use (i assume this impacts xlate failures)
[+] Release apr-1.3.3
[+] Release apr-util-1.3.3
hpux/parisc: all tests pass
hpux/ia64:
testxlate : -Line 63: expected <0>, but saw <22> (non-regression)
Note: Needed to remove __hpux__ stanza in testdso.c (libtool
creates .so not .sl) (non-regression)
aix/ppc32:
testxlate : |Segmentation fault (core dumped) (non-regression)
testsock : |Line 257: expected <0.0.0.0>, but saw
<::ffff:0.0.0.0> (non-regression)
solaris/em64t:
testxlate : |Line 63: expected <0>, but saw <22> (non-regression)
testsock : |Line 92: Problem getting http service (2): No
such file or directory (non-regression)
I didn't have much luck analyzing the AIX crash, in two different
environments dbx complained at length about internal errors in
"stabstring."
--
Eric Covener
covener@gmail.com
Re: 1.3.3 tarballs for review
Posted by Jeff Trawick <tr...@gmail.com>.
On Sat, Aug 9, 2008 at 10:37 AM, William A. Rowe, Jr.
<wr...@rowe-clan.net>wrote:
> Available in the usual; http://apr.apache.org/dev/dist/, win32
> with exported .mak files to follow as soon as I have a bit of
> free time, busy weekend here so likely on Sunday eve (when I'll
> look also at committing Carlo's efforts on MinGW).
>
> +/-1
[+1] Release apr-1.3.3
(tested on OpenSolaris, reviewed diffs)
moot: Release apr-util-1.3.3
Re: 1.3.3 tarballs for review
Posted by Stefan Fritsch <sf...@sfritsch.de>.
On Saturday 09 August 2008, William A. Rowe, Jr. wrote:
> Available in the usual; http://apr.apache.org/dev/dist/, win32
> with exported .mak files to follow as soon as I have a bit of
> free time, busy weekend here so likely on Sunday eve (when I'll
> look also at committing Carlo's efforts on MinGW).
>
> +/-1
> [+1] Release apr-1.3.3
> [ ] Release apr-util-1.3.3
apr 1.3.3 builds fine on Debian/unstable on powerpc and passes all
tests (including testrand). APR_IS_BIGENDIAN is set to 1, as it
should.
Cheers,
Stefan
Re: 1.3.3 tarballs for review
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
William A. Rowe, Jr. wrote:
> Available in the usual; http://apr.apache.org/dev/dist/, win32
> with exported .mak files to follow as soon as I have a bit of
> free time, busy weekend here so likely on Sunday eve (when I'll
> look also at committing Carlo's efforts on MinGW).
>
> +/-1
> [ ] Release apr-1.3.3
Passes; announcement will wait until we can announce both.
> [ ] Release apr-util-1.3.3
Failed/withdrawn due to regression. Next vote coming here in 30 seconds.
Re: 1.3.3 tarballs for review
Posted by Bojan Smojver <bo...@rexursive.com>.
On Sat, 2008-08-09 at 09:37 -0500, William A. Rowe, Jr. wrote:
> +/-1
> [+1] Release apr-1.3.3
> [+1] Release apr-util-1.3.3
Good signatures, good checksums.
Fedora 9, i386/x86_64, all tests passed.
RHEL 4, i386, all tests passed.
CentOS 5, i386/x86_64, all tests passed.
HP-UX 11.11, PA-RISC, 32-bit, the following tests failed (FYI only):
testatomic : -Line 280: Failed creating threads
FAILED 1 of 19
Failed Tests Total Fail Failed %
===================================================
testatomic 19 1 5.26%
testsockets : |Line 70: expected <0>, but saw <221>
|Line 85: expected <0>, but saw <221>
|Line 112: expected <0>, but saw <221>
\Line 185: Condition is false, but expected true
FAILED 3 of 7
Failed Tests Total Fail Failed %
===================================================
testsockets 7 3 42.86%
--
Bojan
Re: 1.3.3 tarballs for review
Posted by Bojan Smojver <bo...@rexursive.com>.
On Tue, 2008-08-12 at 18:22 -0500, William A. Rowe, Jr. wrote:
> That's a pretty major headache,
Yeah, pretty embarrassing, actually :-(
> and if this is the only delta between
> 1.3.3 and 1.3.4, I'd permit an abbreviated vote of 36 hours to release
> 1.3.4 apr-util so we can announce both at once.
Sounds like a good plan to me.
--
Bojan
Re: 1.3.3 tarballs for review
Posted by Olivier Beaton <ol...@quadir.net>.
Perhaps I wasn't clear. Don't get me wrong, I greatly appreciate all
the volunteer work everyone puts into these libraries so that
developers can more efficiently build their applications. I sure
can't resolve these kinds of issues (or I would have and submitted my
own patch) so I appreciate the complexity going on here..
I wasn't griping about "when is the next version going to be out".
I'm talking about the fact that previous releases broke compilation on
a platform (mingw) and stayed that way through multiple releases. In
my case I'm trying to use log4cxx in an enterprise environment and
their hands are tied when a dependant library breaks like this.
I'm told that not this release, but another releasing coming "shortly"
afterwards is going to fix this, but this really seemed like a very
long time to wait for this kind of huge issue to get resolved. Of
course griping isn't going to make it go faster, but I hope it will
make volunteers rank these kinds of issues higher on their list of
priorities in the future because this is an issue many of us really
care about.
Thanks for all the work,
Olivier Beaton
On 8/13/08, Bojan Smojver <bo...@rexursive.com> wrote:
> On Wed, 2008-08-13 at 12:59 -0400, Olivier Beaton wrote:
> > After a nearly 2 month wait for the latest release and svn
> > trunk to compile I've lost a lot of faith.
>
>
> You have to understand that people working on APR/APU a volunteers.
>
> --
>
> Bojan
>
>
Re: 1.3.3 tarballs for review
Posted by Bojan Smojver <bo...@rexursive.com>.
On Wed, 2008-08-13 at 21:51 -0400, Bob Rossi wrote:
> Sure, but what should we do, isn't one of the goals of apr to be
> portable?
It is. Unfortunately, not everyone has or is familiar with all
platforms. For instance, I have no idea what mingw actually is, I never
used it or compiled anything on it. So, it may be a while before you get
the right committer to look over your stuff and test it.
--
Bojan
Re: 1.3.3 tarballs for review
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Bob Rossi wrote:
> On Thu, Aug 14, 2008 at 08:22:49AM +1000, Bojan Smojver wrote:
>> On Wed, 2008-08-13 at 12:59 -0400, Olivier Beaton wrote:
>>> After a nearly 2 month wait for the latest release and svn
>>> trunk to compile I've lost a lot of faith.
>> You have to understand that people working on APR/APU a volunteers.
>
> Sure, but what should we do, isn't one of the goals of apr to be
> portable?
>
> Should I do a binary-search build to see when it broke, and report the
> patch that broke it?
We are happy to move forwards and apply your fix (with all due caution so
that we don't introduce more new regressions!) but it's a matter of one
of the very few committers reviewing it who feel comfortable with the
entire autofoo mess. Unfortunately with autoconf, a positive change for
one platform is all too often breakage to another platform.
But yes, please test frequently and holler at the committer when their patch
breaks your platforms!
Bill
Re: 1.3.3 tarballs for review
Posted by Bob Rossi <bo...@cox.net>.
On Thu, Aug 14, 2008 at 08:22:49AM +1000, Bojan Smojver wrote:
> On Wed, 2008-08-13 at 12:59 -0400, Olivier Beaton wrote:
> > After a nearly 2 month wait for the latest release and svn
> > trunk to compile I've lost a lot of faith.
>
> You have to understand that people working on APR/APU a volunteers.
Sure, but what should we do, isn't one of the goals of apr to be
portable?
Should I do a binary-search build to see when it broke, and report the
patch that broke it?
Thanks,
Bob Rossi
Re: 1.3.3 tarballs for review
Posted by Bojan Smojver <bo...@rexursive.com>.
On Wed, 2008-08-13 at 12:59 -0400, Olivier Beaton wrote:
> After a nearly 2 month wait for the latest release and svn
> trunk to compile I've lost a lot of faith.
You have to understand that people working on APR/APU a volunteers.
--
Bojan
Re: 1.3.3 tarballs for review
Posted by Olivier Beaton <ol...@quadir.net>.
On Tue, Aug 12, 2008 at 7:22 PM, William A. Rowe, Jr.
<wr...@rowe-clan.net> wrote:
> Will tag later tonight, good catch guys.
Is MinGW support still upcoming? I've been waiting for that since I
reported it on June 25th (log4cxx) and June 26th (by Bob Rossi on
apr). After a nearly 2 month wait for the latest release and svn
trunk to compile I've lost a lot of faith.
Olivier Beaton
Re: 1.3.3 tarballs for review
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Bojan Smojver wrote:
> On Tue, 2008-08-12 at 11:11 -0400, Jeff Trawick wrote:
>
>> That free() must be inside a check for APR_HAS_THREADS (potentially
>> extremely painful to diagnose, but for a minority of
>> platforms/builds).
>
> That's my bad guys :-( Thanks for catching it. And thanks Ruediger for
> fixing it.
>
> I hereby change my vote to -1 on apr-util. We can't have things like
> this in the release. We should roll 1.3.4.
That's a pretty major headache, and if this is the only delta between
1.3.3 and 1.3.4, I'd permit an abbreviated vote of 36 hours to release
1.3.4 apr-util so we can announce both at once.
Will tag later tonight, good catch guys.
Re: 1.3.3 tarballs for review
Posted by Bojan Smojver <bo...@rexursive.com>.
On Tue, 2008-08-12 at 11:11 -0400, Jeff Trawick wrote:
> That free() must be inside a check for APR_HAS_THREADS (potentially
> extremely painful to diagnose, but for a minority of
> platforms/builds).
That's my bad guys :-( Thanks for catching it. And thanks Ruediger for
fixing it.
I hereby change my vote to -1 on apr-util. We can't have things like
this in the release. We should roll 1.3.4.
--
Bojan
Re: 1.3.3 tarballs for review
Posted by Ruediger Pluem <rp...@apache.org>.
On 08/12/2008 05:11 PM, Jeff Trawick wrote:
> apr-util:
> testxlate : |Line 63: expected <0>, but saw <22>
> (perhaps the lack of expected charset aliases in Solaris iconv()?)
>
> I guess these last two messages are not errors::
> Failed to load driver file apr_dbd_sqlite2.so
> Failed to load driver file apr_dbd_sqlite3.so
Yep. These are no errors. They simply happen on systems without sqlite installed.
>
> A late code review found this disturbance:
>
> Index: memcache/apr_memcache.c
> ===================================================================
> --- memcache/apr_memcache.c (.../1.3.2) (revision 685172)
> +++ memcache/apr_memcache.c (.../1.3.3) (revision 685172)
> ...
> static apr_status_t
> mc_conn_construct(void **conn_, void *params, apr_pool_t *pool)
> {
> @@ -310,7 +331,11 @@
> return rv;
> }
>
> +#if APR_HAS_THREADS
> + conn = malloc(sizeof( apr_memcache_conn_t )); /* non-pool space! */
> +#else
> conn = apr_palloc(np, sizeof( apr_memcache_conn_t ));
> +#endif
>
> conn->p = np;
>
> @@ -334,34 +359,31 @@
> rv = conn_connect(conn);
> if (rv != APR_SUCCESS) {
> apr_pool_destroy(np);
> + free(conn);
>
> That free() must be inside a check for APR_HAS_THREADS (potentially
> extremely painful to diagnose, but for a minority of platforms/builds).
Thanks for pointing out. Fixed in r685262.
Regards
RĂ¼diger
Re: 1.3.3 tarballs for review
Posted by Jeff Trawick <tr...@gmail.com>.
On Sat, Aug 9, 2008 at 10:37 AM, William A. Rowe, Jr.
<wr...@rowe-clan.net>wrote:
> Available in the usual; http://apr.apache.org/dev/dist/, win32
> with exported .mak files to follow as soon as I have a bit of
> free time, busy weekend here so likely on Sunday eve (when I'll
> look also at committing Carlo's efforts on MinGW).
>
> +/-1
> [ ] Release apr-1.3.3
> [ ] Release apr-util-1.3.3
>
> apr and apr-util test results for 1.3.2 and 1.3.3 match on OpenSolaris:
apr:
testsock : |Line 92: Problem getting http service (2): No such
file or directory
apr-util:
testxlate : |Line 63: expected <0>, but saw <22>
(perhaps the lack of expected charset aliases in Solaris iconv()?)
I guess these last two messages are not errors::
Failed to load driver file apr_dbd_sqlite2.so
Failed to load driver file apr_dbd_sqlite3.so
A late code review found this disturbance:
Index: memcache/apr_memcache.c
===================================================================
--- memcache/apr_memcache.c (.../1.3.2) (revision 685172)
+++ memcache/apr_memcache.c (.../1.3.3) (revision 685172)
...
static apr_status_t
mc_conn_construct(void **conn_, void *params, apr_pool_t *pool)
{
@@ -310,7 +331,11 @@
return rv;
}
+#if APR_HAS_THREADS
+ conn = malloc(sizeof( apr_memcache_conn_t )); /* non-pool space! */
+#else
conn = apr_palloc(np, sizeof( apr_memcache_conn_t ));
+#endif
conn->p = np;
@@ -334,34 +359,31 @@
rv = conn_connect(conn);
if (rv != APR_SUCCESS) {
apr_pool_destroy(np);
+ free(conn);
That free() must be inside a check for APR_HAS_THREADS (potentially
extremely painful to diagnose, but for a minority of platforms/builds).