Re: A push for 2.4.2
There's just 1 more backport left in the list, and its missing just 1 more +1... On Mar 14, 2012, at 3:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now.
Re: A push for 2.4.2
- Original Message - There's just 1 more backport left in the list, and its missing just 1 more +1... Done On Mar 14, 2012, at 3:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. -- Igor Galić Tel: +43 (0) 664 886 22 883 Mail: i.ga...@brainsware.org URL: http://brainsware.org/ GPG: 6880 4155 74BD FD7C B515 2EA5 4B1D 9E08 A097 C9AE
Re: A push for 2.4.2
Thx! And applied! On Apr 3, 2012, at 9:14 AM, Igor Galić wrote: - Original Message - There's just 1 more backport left in the list, and its missing just 1 more +1... Done On Mar 14, 2012, at 3:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. -- Igor Galić Tel: +43 (0) 664 886 22 883 Mail: i.ga...@brainsware.org URL: http://brainsware.org/ GPG: 6880 4155 74BD FD7C B515 2EA5 4B1D 9E08 A097 C9AE
Re: A push for 2.4.2
On 31 Mar 2012, at 3:10 AM, Gregg Smith wrote:
I found this to be an interesting error message:
[Fri Mar 30 18:07:41.019600 2012] [session_crypto:error] [pid 4236:tid 700]
(15)Error string not specified yet: AH01845: (null)
very informative :)
Can you give more details of the crypto driver you're trying to use?
Using the error code AH01845, a quick search of the source shows up this:
rv = apr_crypto_get_driver(driver, conf-library, conf-params, err,
p);
...
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
%s, err-msg);
return rv;
}
It looks like the error code returned by apr_crypto_get_driver() is not
recognised as an APR error, and the error message being printed is the string
(null).
This comes from here:
apr_dso_error(dso, buffer, ERROR_SIZE - 1);
err-msg = buffer;
err-reason = modname;
What this means is that an attempt to load a DSO is failing, with an unknown
APR error code and an error message of (null).
The assumption that the underlying driver will always return a sensible error
message seems to be bogus, this won't help with the message, but will in theory
give a hint to the driver involved:
Index: modules/session/mod_session_crypto.c
===
--- modules/session/mod_session_crypto.c(revision 1307617)
+++ modules/session/mod_session_crypto.c(working copy)
@@ -435,7 +435,7 @@
}
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
-%s, err-msg);
+The crypto library '%s' could not be loaded: %s (%s),
conf-library, err-msg, err-reason);
return rv;
}
if (APR_ENOTIMPL == rv) {
Can you confirm?
Regards,
Graham
--
smime.p7s
Description: S/MIME cryptographic signature
Re: A push for 2.4.2
FYI: needs to be built separately -- httpd-2.4.2 (patched for AIX) builds
and passes configtest on AIX 4.3.3
On Sat, Mar 31, 2012 at 2:33 PM, Graham Leggett minf...@sharp.fm wrote:
On 31 Mar 2012, at 3:10 AM, Gregg Smith wrote:
I found this to be an interesting error message:
[Fri Mar 30 18:07:41.019600 2012] [session_crypto:error] [pid 4236:tid
700] (15)Error string not specified yet: AH01845: (null)
very informative :)
Can you give more details of the crypto driver you're trying to use?
Using the error code AH01845, a quick search of the source shows up this:
rv = apr_crypto_get_driver(driver, conf-library, conf-params,
err, p);
...
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
%s, err-msg);
return rv;
}
It looks like the error code returned by apr_crypto_get_driver() is not
recognised as an APR error, and the error message being printed is the
string (null).
This comes from here:
apr_dso_error(dso, buffer, ERROR_SIZE - 1);
err-msg = buffer;
err-reason = modname;
What this means is that an attempt to load a DSO is failing, with an
unknown APR error code and an error message of (null).
The assumption that the underlying driver will always return a sensible
error message seems to be bogus, this won't help with the message, but will
in theory give a hint to the driver involved:
Index: modules/session/mod_session_crypto.c
===
--- modules/session/mod_session_crypto.c(revision 1307617)
+++ modules/session/mod_session_crypto.c(working copy)
@@ -435,7 +435,7 @@
}
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
-%s, err-msg);
+The crypto library '%s' could not be loaded: %s
(%s), conf-library, err-msg, err-reason);
return rv;
}
if (APR_ENOTIMPL == rv) {
Can you confirm?
Regards,
Graham
--
Re: A push for 2.4.2
Hi Graham,
I know I should have elaborated more but I was very short on time at
that moment. It was a missing dependency.
It was nss and I was pulling away nss dll files to see just what was
needed and what wasn't. Interesting was that the error was different for
most but that error came up for some. The other was
[Fri Mar 30 18:12:05.018600 2012] [session_crypto:error] [pid 3404:tid
700] (OS 126)The specified module could not be found. : AH01845: The
specified module could not be found.
This does give better info
[Sat Mar 31 10:15:59.017600 2012] [session_crypto:error] [pid 4932:tid
700] (15)Error string not specified yet: AH01845: The crypto library
'nss' could not be loaded: (null) (Error during 'nss' initialisation)
Maybe however, Error string not specified by the driver yet or
something to that effect? It cannot be incorrectly read as we're not
sure what to say here yet, as I embarrassingly did.
Regards,
Gregg
On 3/31/2012 5:33 AM, Graham Leggett wrote:
On 31 Mar 2012, at 3:10 AM, Gregg Smith wrote:
I found this to be an interesting error message:
[Fri Mar 30 18:07:41.019600 2012] [session_crypto:error] [pid 4236:tid 700]
(15)Error string not specified yet: AH01845: (null)
very informative :)
Can you give more details of the crypto driver you're trying to use?
Using the error code AH01845, a quick search of the source shows up this:
rv = apr_crypto_get_driver(driver, conf-library, conf-params,err,
p);
...
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
%s, err-msg);
return rv;
}
It looks like the error code returned by apr_crypto_get_driver() is not recognised as an
APR error, and the error message being printed is the string (null).
This comes from here:
apr_dso_error(dso, buffer, ERROR_SIZE - 1);
err-msg = buffer;
err-reason = modname;
What this means is that an attempt to load a DSO is failing, with an unknown APR error
code and an error message of (null).
The assumption that the underlying driver will always return a sensible error
message seems to be bogus, this won't help with the message, but will in theory
give a hint to the driver involved:
Index: modules/session/mod_session_crypto.c
===
--- modules/session/mod_session_crypto.c(revision 1307617)
+++ modules/session/mod_session_crypto.c(working copy)
@@ -435,7 +435,7 @@
}
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
-%s, err-msg);
+The crypto library '%s' could not be loaded: %s (%s),
conf-library, err-msg, err-reason);
return rv;
}
if (APR_ENOTIMPL == rv) {
Can you confirm?
Regards,
Graham
--
Re: A push for 2.4.2
With the code:
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
%s, err-msg);
return rv;
}
then 01845 gets associated with lots of different crypto driver messages.
How about logging something like crypto driver error: %s instead?
--
Tim Bannister – is...@jellybaby.net
Re: A push for 2.4.2
On 31 Mar 2012, at 7:25 PM, Gregg Smith wrote:
It was nss and I was pulling away nss dll files to see just what was needed
and what wasn't. Interesting was that the error was different for most but
that error came up for some. The other was
[Fri Mar 30 18:12:05.018600 2012] [session_crypto:error] [pid 3404:tid 700]
(OS 126)The specified module could not be found. : AH01845: The specified
module could not be found.
This does give better info
[Sat Mar 31 10:15:59.017600 2012] [session_crypto:error] [pid 4932:tid 700]
(15)Error string not specified yet: AH01845: The crypto library 'nss'
could not be loaded: (null) (Error during 'nss' initialisation)
Maybe however, Error string not specified by the driver yet or something to
that effect? It cannot be incorrectly read as we're not sure what to say
here yet, as I embarrassingly did.
The string Error string not specified yet belongs to APR, and is returned if
an error code (in this case
15) isn't recognised by apr_error_string().
If it's nss, then the following code is generating the error:
if (s != SECSuccess) {
if (result) {
apu_err_t *err = apr_pcalloc(pool, sizeof(apu_err_t));
err-rc = PR_GetError();
err-msg = PR_ErrorToName(s);
err-reason = Error during 'nss' initialisation;
*result = err;
}
return APR_ECRYPT;
}
In turn, it means that PR_ErrorToName(s) is returning the string (null).
It may be worth logging the value err-rc as well to see the value of the NSS
error. When I developed this stuff way back when I discovered a number of
places where NSS errors were either missing or were duplicated (ie multiple
failures mapped to the same error code). When the errors were pointed out, the
NSS guys were very quick to fix the problems. Are you using a recent version of
NSS?
Index: modules/session/mod_session_crypto.c
===
--- modules/session/mod_session_crypto.c(revision 1307617)
+++ modules/session/mod_session_crypto.c(working copy)
@@ -435,7 +435,7 @@
}
if (APR_SUCCESS != rv err) {
ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, APLOGNO(01845)
-%s, err-msg);
+The crypto library '%s' could not be loaded: %s (%s:
%d), conf-library, err-msg, err-reason, err-rc);
return rv;
}
if (APR_ENOTIMPL == rv) {
Regards,
Graham
--
smime.p7s
Description: S/MIME cryptographic signature
Re: A push for 2.4.2
On 3/31/2012 12:44 PM, Graham Leggett wrote:
On 31 Mar 2012, at 7:25 PM, Gregg Smith wrote:
It was nss and I was pulling away nss dll files to see just what was needed and
what wasn't. Interesting was that the error was different for most but that
error came up for some. The other was
[Fri Mar 30 18:12:05.018600 2012] [session_crypto:error] [pid 3404:tid 700] (OS
126)The specified module could not be found. : AH01845: The specified module
could not be found.
This does give better info
[Sat Mar 31 10:15:59.017600 2012] [session_crypto:error] [pid 4932:tid 700]
(15)Error string not specified yet: AH01845: The crypto library 'nss' could
not be loaded: (null) (Error during 'nss' initialisation)
Maybe however, Error string not specified by the driver yet or something to that
effect? It cannot be incorrectly read as we're not sure what to say here yet, as I
embarrassingly did.
The string Error string not specified yet belongs to APR, and is returned if
an error code (in this case
15) isn't recognised by apr_error_string().
If it's nss, then the following code is generating the error:
if (s != SECSuccess) {
if (result) {
apu_err_t *err = apr_pcalloc(pool, sizeof(apu_err_t));
err-rc = PR_GetError();
err-msg = PR_ErrorToName(s);
err-reason = Error during 'nss' initialisation;
*result = err;
}
return APR_ECRYPT;
}
In turn, it means that PR_ErrorToName(s) is returning the string (null).
It may be worth logging the value err-rc as well to see the value of the NSS
error. When I developed this stuff way back when I discovered a number of places
where NSS errors were either missing or were duplicated (ie multiple failures
mapped to the same error code). When the errors were pointed out, the NSS guys
were very quick to fix the problems. Are you using a recent version of NSS?
3.13.3, Feb. 22, 2012
Regards,
Gregg
Re: A push for 2.4.2
p.s. just checked the apr-util-2.5.x/xml/expat/README - and it is the same README file. On Thu, Mar 29, 2012 at 9:52 PM, Michael Felt mamf...@gmail.com wrote: Looking for expat I see both in svn for apr-util as well as in the tarball released for public viewing to following info in the projectname/xml/expat/README file: Expat, Release 1.95.7 This is Expat, a C library for parsing XML, written by James Clark. Expat is a stream-oriented XML parser. This means that you register ... Does not look like the new version is being used by apr-util. Or am I still not making any sense? ;) In other words, where should I find the new version of expat to test (build) against? On Thu, Mar 29, 2012 at 9:56 AM, Graham Leggett minf...@sharp.fm wrote: On 25 Mar 2012, at 7:09 PM, Jim Jagielski wrote: How does the week of April 2nd sound? This should provide enough time for the proposed backports to get enough votes and to propose the backports based on recent trunk improvements... Sound like a plan? I volunteer to RM. Quick reminder that apr-util v1.4.2 is up for vote, and would fix some issues for httpd people. It would be good to get the apr-util vote concluded before httpd v2.4.2 goes out the door. Regards, Graham --
Re: A push for 2.4.2
You can choose whatever external expat you like when you configure apr-util.
Re: A push for 2.4.2
OK. Thanks.
I was curious - so I tried to build 2.4.2 on AIX 4.3.3 (still use that
version of AIX to host http://rootvg.net - I do not own it, just help) but
ran into a problem.
apr/apr-util 1.5.0 compiled and installed fine, as did pcre (8.30 I
believe). However, have a problem in the support area...
(ld): lib /usr/lib/librtl.a
LIBRARY: Shared object libaprutil.so: 280 symbols imported.
LIBRARY: Shared object libexpat.a[libexpat.so.0]: 77 symbols imported.
LIBRARY: Shared object libiconv.a[shr4.o]: 10 symbols imported.
LIBRARY: Shared object libiconv.a[shr.o]: 11 symbols imported.
LIBRARY: Shared object libapr.so: 471 symbols imported.
LIBRARY: Shared object libpthread.a[shr_comm.o]: 109 symbols imported.
LIBRARY: Shared object libpthread.a[shr_xpg5.o]: 123 symbols imported.
LIBRARY: Shared object libc_r.a[shr.o]: 2269 symbols imported.
LIBRARY: Shared object libc_r.a[meth.o]: 2 symbols imported.
LIBRARY: Shared object libc_r.a[aio.o]: 11 symbols imported.
LIBRARY: Shared object libc_r.a[pse.o]: 78 symbols imported.
LIBRARY: Shared object libc_r.a[dl.o]: 4 symbols imported.
LIBRARY: Shared object libc_r.a[pty.o]: 1 symbols imported.
LIBRARY: Shared object librtl.a[shr.o]: 1 symbols imported.
LIBRARY: Shared object librtl.a[lazy42.o]: 3 symbols imported.
FILELIST: Number of previously inserted files processed: 13
(ld): exports /usr/lib/libg.exp
EXPORTS: Symbols exported: 4
(ld): initfini _GLOBAL__FI_htpasswd _GLOBAL__FD_htpasswd
(ld): keep XML_Parse
(ld): resolve
RESOLVE: 274 of 4655 symbols were kept.
(ld): addgl /usr/lib/glink.o
ADDGL: Glink code added for 43 symbols.
(ld): er full
ld: 0711-318 ERROR: Undefined symbols were found.
The following symbols are in error:
SymbolInpndx TY CL Source-File(Object-File) OR
Import-File{Shared-object}
RLD: Address Section Rld-type Referencing
Symbol
--
ld: 0711-317 ERROR: Undefined symbol: .apr_generate_random_bytes
.apr_generate_random_bytes [46]ER PR htpasswd.c(htpasswd.o)
001c .textR_RBR[527]
.seed_rand
ER: The return code is 8.
collect2: ld returned 8 exit status
Suggestions?
On Fri, Mar 30, 2012 at 1:46 PM, Eric Covener cove...@gmail.com wrote:
You can choose whatever external expat you like when you configure
apr-util.
Re: A push for 2.4.2
I was curious - so I tried to build 2.4.2 on AIX 4.3.3 (still use that version of AIX to host http://rootvg.net - I do not own it, just help) but ran into a problem. There is not yet a 2.4.2, and your issue is probably best served by creating a dedicated thread rather than using this one intended to prep for 2.4.2, until/unless you think it's a new blocker.
Re: A push for 2.4.2
it is just 2.4.2 on aix 4.3.3 - not real important I expect. Worthy of a mention at least. No to find where this routine is defined. Try again with apr-1.4.x then, On Fri, Mar 30, 2012 at 9:06 PM, Eric Covener cove...@gmail.com wrote: I was curious - so I tried to build 2.4.2 on AIX 4.3.3 (still use that version of AIX to host http://rootvg.net - I do not own it, just help) but ran into a problem. There is not yet a 2.4.2, and your issue is probably best served by creating a dedicated thread rather than using this one intended to prep for 2.4.2, until/unless you think it's a new blocker.
Re: A push for 2.4.2
I found this to be an interesting error message: [Fri Mar 30 18:07:41.019600 2012] [session_crypto:error] [pid 4236:tid 700] (15)Error string not specified yet: AH01845: (null) very informative :) Gregg
Re: A push for 2.4.2
On 25 Mar 2012, at 7:09 PM, Jim Jagielski wrote: How does the week of April 2nd sound? This should provide enough time for the proposed backports to get enough votes and to propose the backports based on recent trunk improvements... Sound like a plan? I volunteer to RM. Quick reminder that apr-util v1.4.2 is up for vote, and would fix some issues for httpd people. It would be good to get the apr-util vote concluded before httpd v2.4.2 goes out the door. Regards, Graham -- smime.p7s Description: S/MIME cryptographic signature
Re: A push for 2.4.2
Looking for expat I see both in svn for apr-util as well as in the tarball released for public viewing to following info in the projectname/xml/expat/README file: Expat, Release 1.95.7 This is Expat, a C library for parsing XML, written by James Clark. Expat is a stream-oriented XML parser. This means that you register ... Does not look like the new version is being used by apr-util. Or am I still not making any sense? ;) In other words, where should I find the new version of expat to test (build) against? On Thu, Mar 29, 2012 at 9:56 AM, Graham Leggett minf...@sharp.fm wrote: On 25 Mar 2012, at 7:09 PM, Jim Jagielski wrote: How does the week of April 2nd sound? This should provide enough time for the proposed backports to get enough votes and to propose the backports based on recent trunk improvements... Sound like a plan? I volunteer to RM. Quick reminder that apr-util v1.4.2 is up for vote, and would fix some issues for httpd people. It would be good to get the apr-util vote concluded before httpd v2.4.2 goes out the door. Regards, Graham --
Re: A push for 2.4.2
Thank you for the compliment. Quite correct I am getting used to svn and where all of you are. For the record, I was able to package the httpd-2.4.x using the apr-1.4.6 package generally available together with the apr-util-1.4.2 tarball in Tarballs/zipballs are at http://apr.apache.org/dev/dist/autoconf-2.68+libtool-2.4.2/. So, probably I am trying to mix the wrong version of apr with an ancient apr-util. Thank you all for the suggestions and corrections. On Mon, Mar 26, 2012 at 11:14 PM, William A. Rowe Jr. wr...@rowe-clan.netwrote: On 3/25/2012 5:18 PM, Michael Felt wrote: I have been trying to build trunk from trunk versions of apr and apr-util. buildconf complains about not being able to find APR-util (or apr-UTIL). In any case, caps are involved. Windows might not complain, but UNIX does. You make no sense. The trunk version of apr 2.0.0-dev includes all '-util' functionality. In fact, there is no apr-util 'trunk' anymore. What you might want for development are the current apr-util and apr 1.x branches, 1.4.x are current, and 1.5.x are 'next' if we get that far. But don't mix 2.0 apr with an apr-util at all. The merged into apr.
Re: A push for 2.4.2
A new expat 2.1.0 is available, like to see it also included in apr and test with 2.4.2. Bill was waiting for it: From: William A. Rowe Jr. Date: 2012-03-08 20:27:25 Nor am I (aware of issues), however expat 2.0.1 has vulnerabilites which are corrected in the expat project's svn but not in a 'release' (we have a patched flavor of 1.9.5 in apr project's repository). We won't ship the apr 1.4.6 +patch, but would wait for apr to release again. OpenSSL is about to deliver 1.0.1 (and then deprecate 1.0.0 long before we finish any httpd 2.4 cycle) so waiting the additional week or few for the brand new openssl 1.0.1 and a new apr tag (and a new expat 2.0.2 would really be lovely). -Original Message- From: Jim Jagielski Sent: Wednesday, March 14, 2012 8:36 PM Newsgroups: gmane.comp.apache.devel To: dev@httpd.apache.org Subject: A push for 2.4.2 In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now.
Re: A push for 2.4.2
On 3/27/2012 7:22 AM, Steffen wrote: A new expat 2.1.0 is available, like to see it also included in apr and test with 2.4.2. Excellent news! Of course OpenSSL 1.0.1 has since been released. I'm likely to get to a quick test build Thursday, so that if there is breakage we can do something about it, before Jim's TR next week.
Re: A push for 2.4.2
On 3/25/2012 5:18 PM, Michael Felt wrote: I have been trying to build trunk from trunk versions of apr and apr-util. buildconf complains about not being able to find APR-util (or apr-UTIL). In any case, caps are involved. Windows might not complain, but UNIX does. You make no sense. The trunk version of apr 2.0.0-dev includes all '-util' functionality. In fact, there is no apr-util 'trunk' anymore. What you might want for development are the current apr-util and apr 1.x branches, 1.4.x are current, and 1.5.x are 'next' if we get that far. But don't mix 2.0 apr with an apr-util at all. The merged into apr.
Re: A push for 2.4.2
How does the week of April 2nd sound? This should provide enough time for the proposed backports to get enough votes and to propose the backports based on recent trunk improvements... Sound like a plan? I volunteer to RM. On Mar 14, 2012, at 3:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now.
Re: A push for 2.4.2
On 25.03.2012 19:09, Jim Jagielski wrote: How does the week of April 2nd sound? This should provide enough time for the proposed backports to get enough votes and to propose the backports based on recent trunk improvements... Sound like a plan? I volunteer to RM. On Mar 14, 2012, at 3:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. +1 Rainer
Re: A push for 2.4.2
On 25 Mar 2012, at 7:09 PM, Jim Jagielski wrote: How does the week of April 2nd sound? This should provide enough time for the proposed backports to get enough votes and to propose the backports based on recent trunk improvements... Sound like a plan? I volunteer to RM. +1. I've just TR'ed apr-util v1.4.2, containing some compile time fixes for static builds. In theory, given no drama it should be ready for April 2nd. Regards, Graham -- smime.p7s Description: S/MIME cryptographic signature
Re: A push for 2.4.2
I have been trying to build trunk from trunk versions of apr and apr-util. buildconf complains about not being able to find APR-util (or apr-UTIL). In any case, caps are involved. Windows might not complain, but UNIX does. To test what you are testing - should I use apr and apr-util trunks, or is it better in all thinks - i.e. including httpd-trunk builds to use the released versions of apr, or their trunk versions. Reading the lists I see many of you are involved in both projects. I just want to be in sync with the ways you test for new releases. Sincerely, Michael On Sun, Mar 25, 2012 at 10:40 PM, Graham Leggett minf...@sharp.fm wrote: On 25 Mar 2012, at 7:09 PM, Jim Jagielski wrote: How does the week of April 2nd sound? This should provide enough time for the proposed backports to get enough votes and to propose the backports based on recent trunk improvements... Sound like a plan? I volunteer to RM. +1. I've just TR'ed apr-util v1.4.2, containing some compile time fixes for static builds. In theory, given no drama it should be ready for April 2nd. Regards, Graham --
Re: A push for 2.4.2
On Sun, Mar 25, 2012 at 6:18 PM, Michael Felt mamf...@gmail.com wrote: I have been trying to build trunk from trunk versions of apr and apr-util. buildconf complains about not being able to find APR-util (or apr-UTIL). In any case, caps are involved. Windows might not complain, but UNIX does. To test what you are testing - should I use apr and apr-util trunks, or is it better in all thinks - i.e. including httpd-trunk builds to use the released versions of apr, or their trunk versions. Reading the lists I see many of you are involved in both projects. I just want to be in sync with the ways you test for new releases. use apr 1.4.x and apr-util 1.4.x with httpd trunk or httpd 2.4.x
Re: A push for 2.4.2
On 26 Mar 2012, at 12:18 AM, Michael Felt wrote:
I have been trying to build trunk from trunk versions of apr and apr-util.
buildconf complains about not being able to find APR-util (or apr-UTIL). In
any case, caps are involved. Windows might not complain, but UNIX does.
In theory, you should be able to get away with not calling buildconf at all,
given that this command is run before the tarballs are created and it should
all be functional from the outset. Or to put it another way, you only need to
run buildconf if you checked the code out from svn, instead of trying to build
from a tarball.
The ./configure script needs the base path of the apr and apr-util
installations passed to it, something like this:
./configure [options] --with-apr=%{_prefix} --with-apr-util=%{_prefix}
where prefix is /opt/local (for you, as I recall).
Regards,
Graham
--
smime.p7s
Description: S/MIME cryptographic signature
Re: A push for 2.4.2
On 15/03/2012 07:29, William A. Rowe Jr. wrote: On 3/14/2012 2:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. Then this afternoon I'll propose a 'really small patch' around 'the win32 issue' and ask folks who have had problems, and those who have not, to test it out. It appears MSDN documentation was not our friend and their directives were misplaced, and the patch should drop some code (not used back in 2.2) which might be the underlying cause of Steffan and crew's issues. We would probably hear back within the day, so windows folks would be very happy with a TR sometime Friday-Monday. I just can't recall offhand if all the win32 quirks in apr were resolved in 1.4.6, or a new TR is still needed on that side bill++ :) Any chance you can elaborate on It appears MSDN documentation was not our friend... etc? Issac
Re: A push for 2.4.2
On 15/03/2012 07:29, William A. Rowe Jr. wrote: On 3/14/2012 2:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. Then this afternoon I'll propose a 'really small patch' around 'the win32 issue' and ask folks who have had problems, and those who have not, to test it out. It appears MSDN documentation was not our friend and their directives were misplaced, and the patch should drop some code (not used back in 2.2) which might be the underlying cause of Steffan and crew's issues. We would probably hear back within the day, so windows folks would be very happy with a TR sometime Friday-Monday. I just can't recall offhand if all the win32 quirks in apr were resolved in 1.4.6, or a new TR is still needed on that side bill++ :) Any chance you can elaborate on It appears MSDN documentation... etc? Issac
Re: A push for 2.4.2
On 3/15/2012 4:07 AM, Issac Goldstand wrote: Any chance you can elaborate on It appears MSDN documentation was not our friend... etc? From MSDN and my understanding of the new wait-on-event API, it appeared that MSDN suggested these would be defaults and we would have to adjust for apr's assumptions (in server/mpm/winnt/child.c); /* Restore the state corresponding to apr_os_sock_make's default * assumption of timeout -1 (really, a flaw of os_sock_make and * os_sock_put that it does not query to determine -timeout). * XXX: Upon a fix to APR, these three statements should disappear. */ ioctlsocket(context-accept_socket, FIONBIO, zero); setsockopt(context-accept_socket, SOL_SOCKET, SO_RCVTIMEO, (char *) zero, sizeof(zero)); setsockopt(context-accept_socket, SOL_SOCKET, SO_SNDTIMEO, (char *) zero, sizeof(zero)); but no, apparently this is not the case, and these can't be assumed to be appropriate for both AcceptEx and accept style logic.
Re: A push for 2.4.2
Bill, Us Windows folk would be ecstatic! It is affecting a few noisy users, then there's the silent masses :) Not sure what APR quirks, refresh my memory. I'm +1 for a APU 1.4.2 also, crypto not building for static lib throws a decent sized monkey wrench into the httpd build. Then maybe we can connect the dots in the .dsw files and in makefile.win (which you have started sort of). Gregg On 3/14/2012 10:29 PM, William A. Rowe Jr. wrote: On 3/14/2012 2:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. Then this afternoon I'll propose a 'really small patch' around 'the win32 issue' and ask folks who have had problems, and those who have not, to test it out. It appears MSDN documentation was not our friend and their directives were misplaced, and the patch should drop some code (not used back in 2.2) which might be the underlying cause of Steffan and crew's issues. We would probably hear back within the day, so windows folks would be very happy with a TR sometime Friday-Monday. I just can't recall offhand if all the win32 quirks in apr were resolved in 1.4.6, or a new TR is still needed on that side?
Re: A push for 2.4.2
Gregg, I believe these overrides are either not needed at all, or are not needed in specific cases, and have yet to determine which is the case that the users are experiencing. Is anyone complaining about AcceptFilter data or connection? If not, then these lines of code simply need to be dodged for AcceptFilter none. On 3/15/2012 6:06 AM, Gregg Smith wrote: Bill, Us Windows folk would be ecstatic! It is affecting a few noisy users, then there's the silent masses :) Not sure what APR quirks, refresh my memory. I'm +1 for a APU 1.4.2 also, crypto not building for static lib throws a decent sized monkey wrench into the httpd build. Then maybe we can connect the dots in the .dsw files and in makefile.win (which you have started sort of). Gregg On 3/14/2012 10:29 PM, William A. Rowe Jr. wrote: On 3/14/2012 2:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. Then this afternoon I'll propose a 'really small patch' around 'the win32 issue' and ask folks who have had problems, and those who have not, to test it out. It appears MSDN documentation was not our friend and their directives were misplaced, and the patch should drop some code (not used back in 2.2) which might be the underlying cause of Steffan and crew's issues. We would probably hear back within the day, so windows folks would be very happy with a TR sometime Friday-Monday. I just can't recall offhand if all the win32 quirks in apr were resolved in 1.4.6, or a new TR is still needed on that side?
Re: A push for 2.4.2
On 15 Mar 2012, at 1:06 PM, Gregg Smith wrote: Us Windows folk would be ecstatic! It is affecting a few noisy users, then there's the silent masses :) Not sure what APR quirks, refresh my memory. I'm +1 for a APU 1.4.2 also, crypto not building for static lib throws a decent sized monkey wrench into the httpd build. Then maybe we can connect the dots in the .dsw files and in makefile.win (which you have started sort of). Happy to RM an apr-util release. Am I right in understanding the issues are fixed and we're good to go? Regards, Graham -- smime.p7s Description: S/MIME cryptographic signature
Re: A push for 2.4.2
On Mar 15, 2012, at 7:06 AM, Gregg Smith wrote: Bill, Us Windows folk would be ecstatic! It is affecting a few noisy users, then there's the silent masses :) Not sure what APR quirks, refresh my memory. I'm +1 for a APU 1.4.2 also, crypto not building for static lib throws a decent sized monkey wrench into the httpd build. Then maybe we can connect the dots in the .dsw files and in makefile.win (which you have started sort of). +1 for APU 1.4.2 (adding APR to this thread)... I can RM if need be.
Re: A push for 2.4.2
Bill, not that pages are not showing up, the only problem with AcceptFilter data has been the AcceptEx error/becoming unresponsive. I admit to never suggesting trying AcceptFilter connect. Of course, the problem is on the https side so I never tell them to go to none for https. I leave mine at the default (data) for https, have a script that monitors the error log and sends a graceful restart when it detects a new AcceptEx error. I do not think I have ever recieved a blank/partial page with the default. If you have a patch, I'll be glad to patch 2.4 head and try it out. On 3/15/2012 4:13 AM, William A. Rowe Jr. wrote: Gregg, I believe these overrides are either not needed at all, or are not needed in specific cases, and have yet to determine which is the case that the users are experiencing. Is anyone complaining about AcceptFilter data or connection? If not, then these lines of code simply need to be dodged for AcceptFilter none. On 3/15/2012 6:06 AM, Gregg Smith wrote: Bill, Us Windows folk would be ecstatic! It is affecting a few noisy users, then there's the silent masses :) Not sure what APR quirks, refresh my memory. I'm +1 for a APU 1.4.2 also, crypto not building for static lib throws a decent sized monkey wrench into the httpd build. Then maybe we can connect the dots in the .dsw files and in makefile.win (which you have started sort of). Gregg On 3/14/2012 10:29 PM, William A. Rowe Jr. wrote: On 3/14/2012 2:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. Then this afternoon I'll propose a 'really small patch' around 'the win32 issue' and ask folks who have had problems, and those who have not, to test it out. It appears MSDN documentation was not our friend and their directives were misplaced, and the patch should drop some code (not used back in 2.2) which might be the underlying cause of Steffan and crew's issues. We would probably hear back within the day, so windows folks would be very happy with a TR sometime Friday-Monday. I just can't recall offhand if all the win32 quirks in apr were resolved in 1.4.6, or a new TR is still needed on that side?
Re: A push for 2.4.2
Graham, Current APU 1.4 head builds fine. It looks like your fix for this is the only thing in changes. On 3/15/2012 4:13 AM, Graham Leggett wrote: On 15 Mar 2012, at 1:06 PM, Gregg Smith wrote: Us Windows folk would be ecstatic! It is affecting a few noisy users, then there's the silent masses :) Not sure what APR quirks, refresh my memory. I'm +1 for a APU 1.4.2 also, crypto not building for static lib throws a decent sized monkey wrench into the httpd build. Then maybe we can connect the dots in the .dsw files and in makefile.win (which you have started sort of). Happy to RM an apr-util release. Am I right in understanding the issues are fixed and we're good to go? Regards, Graham --
Re: A push for 2.4.2
On 3/14/2012 2:36 PM, Jim Jagielski wrote: In the attempts to keep the momentum going, I'd like to push for a httpd 2.4.2 release Real Soon Now. Then this afternoon I'll propose a 'really small patch' around 'the win32 issue' and ask folks who have had problems, and those who have not, to test it out. It appears MSDN documentation was not our friend and their directives were misplaced, and the patch should drop some code (not used back in 2.2) which might be the underlying cause of Steffan and crew's issues. We would probably hear back within the day, so windows folks would be very happy with a TR sometime Friday-Monday. I just can't recall offhand if all the win32 quirks in apr were resolved in 1.4.6, or a new TR is still needed on that side?
