You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@apr.apache.org by David Reid <da...@jetnet.co.uk> on 2004/08/26 10:09:38 UTC
RC6 take 2
new apr-util tarballs are now available at http://www.apache.org/~dreid/
So far I've seen only 1 +1 for a release. Anyone else care to vote?
david
Re: RC6 take 2
Posted by Joe Orton <jo...@redhat.com>.
On Thu, Aug 26, 2004 at 09:09:38AM +0100, David Reid wrote:
> new apr-util tarballs are now available at http://www.apache.org/~dreid/
+1 for these too. Tested using --with-ldap too this time, there are
compiler warnings in the LDAP code but nothing critical.
joe
Re: RC6 take 2
Posted by Cliff Woolley <jw...@virginia.edu>.
the fix for this problem was actually committed to apr-util several months
ago. But for some reason the fix did not make it in to the tag of any of
the RC's. seems like a tagging problem. David, can you just bump the tag
on apr-util/xlate/xlate.c up from version 1.21 to 1.22? Then it will all
be good.
--Cliff
On Thu, 26 Aug 2004, Cliff Woolley wrote:
>
> Has anybody gotten apr-util to compile with apr-iconv?
>
> I get the following on win32 when building apr-util with apr and apr-iconv
> all present:
>
> apr-util\xlate\xlate.c(181): error C2198: 'apr_iconv_close' : too few
> arguments for call through pointer-to-function
> apr-util\xlate\xlate.c(182): error C2198: 'apr_iconv_open' : too few
> arguments for call through pointer-to-function
> apr-util\xlate\xlate.c(182): warning C4047: '=' : 'apr_iconv_t' differs
> in levels of indirection from 'apr_status_t'
>
> It would be a two line fix to xlate.c to get rid of this problem.
> Honestly it seems unlikely anybody's ever even attempted to compile on
> this path, so I'm not sure if I care. And in any event, other than that
> it looks good. +1 with or without a fix.
>
> --Cliff
>
Re: RC6 take 2
Posted by Cliff Woolley <jw...@virginia.edu>.
Has anybody gotten apr-util to compile with apr-iconv?
I get the following on win32 when building apr-util with apr and apr-iconv
all present:
apr-util\xlate\xlate.c(181): error C2198: 'apr_iconv_close' : too few
arguments for call through pointer-to-function
apr-util\xlate\xlate.c(182): error C2198: 'apr_iconv_open' : too few
arguments for call through pointer-to-function
apr-util\xlate\xlate.c(182): warning C4047: '=' : 'apr_iconv_t' differs
in levels of indirection from 'apr_status_t'
It would be a two line fix to xlate.c to get rid of this problem.
Honestly it seems unlikely anybody's ever even attempted to compile on
this path, so I'm not sure if I care. And in any event, other than that
it looks good. +1 with or without a fix.
--Cliff
Re: RC6 take 2
Posted by Mads Toftum <ma...@toftum.dk>.
On Thu, Aug 26, 2004 at 12:29:46PM -0700, Aaron Bannert wrote:
> +1 (seems to work fine on Solaris 9/x86)
>
Success on solaris sparc 8 too with the small detail that I had
to disable ipv6 to avoid a segfault in the tcp test.
vh
Mads Toftum
--
`Darn it, who spiked my coffee with water?!' - lwall
Re: RC6 take 2
Posted by Aaron Bannert <aa...@clove.org>.
+1 (seems to work fine on Solaris 9/x86)
-aaron
On Aug 26, 2004, at 1:09 AM, David Reid wrote:
> new apr-util tarballs are now available at
> http://www.apache.org/~dreid/
>
> So far I've seen only 1 +1 for a release. Anyone else care to vote?
>
> david
>
Re: RC6 take 2
Posted by Mladen Turk <mt...@apache.org>.
Justin Erenkrantz wrote:
>
>> Well, I should run the test with verbose:
>> Listed only the one with verbose messages
>> (both SUCCESS and FAIL).
>
>
> I hate to ask this: but have these tests ever worked? I guess it's
> sucky that some tests fail on Win32, but if they never worked, I don't
> consider it a showstopper. (There just aren't enough Win32 developers
> to look at it?)
>
Well, I'm not an apr commiter.
In about 10 patches I propose eventually
one or two get even looked in :(.
Here is the patch for makefile that at least makes testall not hanging.
(There was no readchild.exe to read the pipes).
So, you may commit that, and one error less :).
Index: Makefile.win
===================================================================
RCS file: /home/cvspublic/apr/test/Makefile.win,v
retrieving revision 1.22
diff -u -r1.22 Makefile.win
--- Makefile.win 26 Jul 2004 15:21:59 -0000 1.22
+++ Makefile.win 26 Aug 2004 16:34:40 -0000
@@ -6,7 +6,8 @@
sendfile.exe \
proc_child.exe \
tryread.exe \
- occhild.exe\
+ occhild.exe \
+ readchild.exe \
sockchild.exe \
testlockperf.exe \
testshmproducer.exe \
@@ -39,6 +40,9 @@
occhild.exe: occhild.obj $(LOCAL_LIBS)
$(LINK) occhild.obj $(LOCAL_LIBS) $(ALL_LIBS)
+
+readchild.exe: readchild.obj $(LOCAL_LIBS)
+ $(LINK) readchild.obj $(LOCAL_LIBS) $(ALL_LIBS)
proc_child.exe: proc_child.obj $(LOCAL_LIBS)
$(LINK) /debug /subsystem:console /machine:I386 \
> We *need* to get 1.0 out the door. We've done *SIX* (to paraphrase our
> British developers, "bloody fricking *SIX*") release candidates for 1.0
> so far. This is insane! I don't blame David if he loses his sanity
> about now.
>
> 1.0 does *not* need to perfect. At some point, we need to ship. -- justin
>
>
Sure.
But please don't shoot the messanger :).
Regards,
MT.
Re: RC6 take 2
Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
--On Thursday, August 26, 2004 6:15 PM +0200 Mladen Turk <mt...@apache.org>
wrote:
> Well, I should run the test with verbose:
> Listed only the one with verbose messages
> (both SUCCESS and FAIL).
I hate to ask this: but have these tests ever worked? I guess it's sucky that
some tests fail on Win32, but if they never worked, I don't consider it a
showstopper. (There just aren't enough Win32 developers to look at it?)
We *need* to get 1.0 out the door. We've done *SIX* (to paraphrase our
British developers, "bloody fricking *SIX*") release candidates for 1.0 so
far. This is insane! I don't blame David if he loses his sanity about now.
1.0 does *not* need to perfect. At some point, we need to ship. -- justin
Re: RC6 take 2
Posted by Mladen Turk <mt...@apache.org>.
David Reid wrote:
> Justin Erenkrantz wrote:
>
>> --On Thursday, August 26, 2004 9:09 AM +0100 David Reid
>> <da...@jetnet.co.uk> wrote:
>>
>>> new apr-util tarballs are now available at http://www.apache.org/~dreid/
>>>
>>> So far I've seen only 1 +1 for a release. Anyone else care to vote?
>>
>>
>>
>> +1. Passes testall for both apr and apr-util on Darwin. -- justin
>
>
> Thanks. Anyone from the world of windows care to comment...
>
Well, I should run the test with verbose:
Listed only the one with verbose messages
(both SUCCESS and FAIL).
testatomic : Line 264: Invalid return value from thread_join
FAILED 1 of 14
testfileinfo : Line 102: apr_stat and apr_getfileinfo differ in
protection
FAILED 1 of 5
testlfs : Line 259: Large Files not supported
SUCCESS
(That's odd, cause there is '#define APR_HAS_LARGE_FILES 1' in apr.hw)
testmmap : Line 86: expected <29>, but saw <28>
Line 103: expected <29>, but saw <28>
Line 127: expected <is the MMAP data file.
>, but saw <is the MMAP data file.
>
FAILED 3 of 8
testprocmutex : Line 141: APR lacks fork() support
SUCCESS
testrand2 : 0x41,0xb5,0x16,0x94,0x66,0x07,0x6a,0x55,
...
0x82,0x04,0xad,0xd2,0x0b,0xd5,0xac,0xda,
...
Line 102: Randomness mismatch
0x05,0xad,0x7c,0x83,0x9a,0xb1,0x9e,0x47,
....
0x38,0x8f,0x01,0x29,0x5a,0x5c,0x1f,0xa8,
...
Line 102: Randomness mismatch
0xfd,0xf1,0xbb,0xb7,0x70,0x78,0x1d,0x1a,
...
0xe8,0xe7,0xc9,0x45,0xe2,0x2a,0x54,0xb2,
...
Line 102: Randomness mismatch
0x26,0x61,0x4e,0x94,0xf8,0xb9,0xba,0x5a,
...
0x7d,0x0e,0xc4,0x4e,0x3e,0xac,0x86,0x50,
...
Line 102: Randomness mismatch
FAILED 4 of 7
testshm : Line 231: Error removing shared memory block
SUCCESS
testsockets : Line 82: IPv6 Line 101: IPv6
SUCCESS
testsockopt : Line 83: TCP isn't corkable
SUCCESS
testuser : Line 63: Groups from apr_uid_get
SUCCESS
Regards,
MT.
Re: RC6 take 2
Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
--On Thursday, August 26, 2004 6:41 PM +0200 Graham Leggett <mi...@sharp.fm>
wrote:
> Test failures are bugs, and bugs can be fixed in v1.0.1 and beyond.
Exactly. -- justin
Re: RC6 take 2
Posted by Graham Leggett <mi...@sharp.fm>.
David Reid wrote:
> If anyone tries to veto the release for these problem they better have
> such a bloody good technical reason that it would get Saddam out of jail
> and back onto the streets...
Test failures are bugs, and bugs can be fixed in v1.0.1 and beyond.
Regards,
Graham
--
Re: RC6 take 2
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
At 11:37 AM 8/26/2004, David Reid wrote:
>Mladen Turk wrote:
>
>>David Reid wrote:
>> >
>> > Thanks. Anyone from the world of windows care to comment...
>> >
>>Here are some windows comments :) .
>
>Mladen - please don't take this personally - it's not directed at you directly.
>There have been 6 RC's and this is the first time that anyone has brought these failures up. I'm sure they've been there all along.
>
>If anyone tries to veto the release for these problem they better have such a bloody good technical reason that it would get Saddam out of jail and back onto the streets...
IIUC - these are not API problems, they are bugs?
Let's keep moving. Only one reason to stop and rethink, either
it doesn't install correctly (or clobbers binary-incompatible APR-0's),
or something is so broken the API must be changed.
None of those are on my radar now.
Mladen - no reason not to fix those bugs, and they will be picked
up in 1.0.1. Nobody is ready to release software based on APR 1.0,
it didn't exist. Once we release, they beta. They will find bugs.
By the time they actually have stable code for APR 1.0, we will
have caught up to 1.0.2 or beyond w/ bug fixes. None of which
should possibly break their code, unless they are relying on some
broken behavior and apr code errors..
As David mentioned, sorry if his response was harsh. If you haven't
followed the progress, it's been, uhm, glacial at best :)
Carry on.
Bill
Re: RC6 take 2
Posted by David Reid <da...@jetnet.co.uk>.
Mladen Turk wrote:
> David Reid wrote:
>
> >
> > Thanks. Anyone from the world of windows care to comment...
> >
>
> Here are some windows comments :) .
Mladen - please don't take this personally - it's not directed at you
directly.
I had to go for a short break and count to ten before replying to this.
I hope it was enough!
At this point I have only a single comment - "so what"?
There have been 6 RC's and this is the first time that anyone has
brought these failures up. I'm sure they've been there all along. Sure,
there have been some patches for windows, but far less has been done
towards improving the code for Windows than other platforms during this
marathon that used to be known as a release process.
If those people out there who develop for windows have been unable or
unwilling to fix these then I'm afraid that at this point I REALLY DON'T
FRICKING CARE!!!!!!
If anyone tries to veto the release for these problem they better have
such a bloody good technical reason that it would get Saddam out of jail
and back onto the streets...
david
Re: RC6 take 2
Posted by Mladen Turk <mt...@apache.org>.
David Reid wrote:
>
> Thanks. Anyone from the world of windows care to comment...
>
Here are some windows comments :) .
Compilation:
sha2.c
random\unix\sha2.c(461) : warning C4244: '=' : conversion from
'apr_uint64_t' to 'unsigned int', possible loss of data
random\unix\sha2.c(507) : warning C4244: '=' : conversion from
'apr_uint64_t' to 'unsigned int', possible loss of data
random\unix\sha2.c(783) : warning C4244: '=' : conversion from
'apr_uint64_t' to 'unsigned int', possible loss of data
random\unix\sha2.c(823) : warning C4244: '=' : conversion from
'apr_uint64_t' to 'unsigned int', possible loss of data
Would be nice to fix that.
Test (WIN2K3) :
testall.exe:
testatomic : FAILED 1 of 14
testdir : SUCCESS
testdso : SUCCESS
testdup : SUCCESS
testenv : SUCCESS
testfile : SUCCESS
testfilecopy : SUCCESS
testfileinfo : FAILED 1 of 5
testflock : SUCCESS
testfmt : SUCCESS
testfnmatch : SUCCESS
testargs : SUCCESS
testhash : SUCCESS
testipsub : SUCCESS
testlock : SUCCESS
testlfs : SUCCESS
testmmap : FAILED 3 of 8
testnames : SUCCESS
testoc : SUCCESS
testpath : SUCCESS
testpipe : /
Blocks on testpipe! Had to comment the entry in abts_tests.h
testpath : SUCCESS
testpoll : SUCCESS
testpools : SUCCESS
testproc : SUCCESS
testprocmutex : SUCCESS
testrand : SUCCESS
testrand2 : -0x41,0xb5,0x16,0x94,0x66,0x0
....
FAILED 4 of 7
testsleep : SUCCESS
testshm : SUCCESS
testsock : SUCCESS
testsockets : SUCCESS
testsockopt : SUCCESS
teststr : SUCCESS
teststrnatcmp : SUCCESS
testtable : SUCCESS
testtemp : SUCCESS
testthread : SUCCESS
testtime : SUCCESS
testud : SUCCESS
testuser : SUCCESS
testvsn : SUCCESS
Failed Tests Total Fail Failed %
===================================================
testatomic 14 1 7.14%
testfileinfo 5 1 20.00%
testmmap 8 3 37.50%
testrand2 7 4 57.14%
Regards,
MT.
Re: RC6 take 2
Posted by David Reid <da...@jetnet.co.uk>.
Justin Erenkrantz wrote:
> --On Thursday, August 26, 2004 9:09 AM +0100 David Reid
> <da...@jetnet.co.uk> wrote:
>
>> new apr-util tarballs are now available at http://www.apache.org/~dreid/
>>
>> So far I've seen only 1 +1 for a release. Anyone else care to vote?
>
>
> +1. Passes testall for both apr and apr-util on Darwin. -- justin
Thanks. Anyone from the world of windows care to comment...
I read that as 3 +1's now. While it's enough it'd be very nice if some
others would chime in as well...
david
Re: RC6 take 2
Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
--On Thursday, August 26, 2004 9:09 AM +0100 David Reid <da...@jetnet.co.uk>
wrote:
> new apr-util tarballs are now available at http://www.apache.org/~dreid/
>
> So far I've seen only 1 +1 for a release. Anyone else care to vote?
+1. Passes testall for both apr and apr-util on Darwin. -- justin
Re: RC6 take 2
Posted by "Brian W. Fitzpatrick" <fi...@red-bean.com>.
> David Reid wrote:
>
> > new apr-util tarballs are now available at http://www.apache.org/~dreid/
> >
> > So far I've seen only 1 +1 for a release. Anyone else care to vote?
+1
-Fitz
Re: RC6 take 2
Posted by Bill Stoddard <bi...@wstoddard.com>.
David Reid wrote:
> new apr-util tarballs are now available at http://www.apache.org/~dreid/
>
> So far I've seen only 1 +1 for a release. Anyone else care to vote?
>
> david
>
+1
bill