You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Jim Jagielski <ji...@jaguNET.com> on 1997/11/16 16:17:05 UTC
[STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Very, very close... At present, it looks like the
only thing is the ``spaces in NT argv'' problem.
Paul has a suggested patch, which Ben agrees with,
but Ben points out that a more general solution
is needed. He will try to work one out. We'll hold
off on 1.3b3 until Ben decides.
==================================================
As of:
Sun Nov 16 10:07:28 EST 1997
Release:
2.0 : In the future
1.3b3: In development... Release 11/16/97 or so
1.3b2: Tarball rolled and available
Source tagged as APACHE_1_3b2
Apache web-page reflects availability of 1.3b2
Announcement released
1.3b1: There is no 1.3b1
Plan:
* At present, we are close to releasing 1.3b3. We are waiting
for a resolution of the "blank in filenames" problem in Win32
before we tag and freeze 1.3b3. There are some patches that
have enough votes to be included in 1.3b3 if commited soon.
When the above mention "problem" is fixed, the 1.3b3 code
will be CVS tagged, but a tarball won't be rolled. This is
so we can roll the UNIX stuff and the Win32 stuff at the same
time, at which point we can formally announce. Ken wants to
do the announcement.
Showstoppers:
Committed:
* Ken's perl logfile splitter (to ./support)
* Lars Eilebrecht's [PATCH] mod_speling loglevels
* Ben's Fix CGI under Win95 (plus some error checking)
* Ben's Fix absolute path detection
* Martin's [PATCH] Reject multiple arguments to <...> block directives
* Martin's [PATCH] Detect missing </Directive> at end-of-file
* PR#1284 fix: add UNIQUE_ID to suexec wrapper for mod_unique_id
* Ken's [PATCH] Slight cleanup of logging of bad directives
* Martin's [PATCH] Suggestion for HTTP_VERSION(maj,min) macro
* PR#1030 fix: Fix problem causing crash when a CGI is accessed with
a ?arg does not contain a =
* Work around buggy stat() on Win95 (closes some PR's?)
* Paul's replacement fo README.NT
* Paul's [BUG] NT fixup for failed spawn of child
* Dean's [PATCH] PR#1241: mod_dir/autoindex should DECLINE PUT
* Dean's [PATCH] mod_speling 64-bit issues
* Martin's [PATCH] Global variables should be set in MainServer,
* Sameer's Fix builds so that they depend on the .o files
* Dean's adjustment of mod_unique_id.c
* Email address added to all internal LICENSEs
* Rasmus's abbreviated [PATCH] Trying to grok .module files
* Sameer's [PATCH] pathnames
* Martin's [PATCH] Take 3: Line Number counting in config file
* PR#849: AIX now uses USE_FCNTL_SERIALIZED_ACCEPT.
* PR#1267: i386 AIX does not have memmove.
* PR#482, 1246: HPUX now defaults to using Spencer regex.
* PR#977: Some versions of netbsd don't automatically define __NetBSD__
* PR#1249: Unixware 2.x requires -lgen for syslog
* All SVR4 systems now use NET_SIZE_T = size_t
* PR#1321: Solaris 2.4 needs spencer regex
* WANTSREGEX=YES now the default
* Paul's Dean's patches and NT, and a PATCH or two
* Dean's [PATCH] ensure config errors/warnings are logged
* Dean's [PATCH] PR#467: serialized accept
* MODULE_MAGIC_NUMBER bumped up
* Dean's [PATCH] PR#1205: mod_mime_magic cleanups
* Dean's [PATCH] PR#1317: AIX 1.x can't have files named @
* Dean's [PATCH] Re: New error_log format (fwd)
* Ben's [PATCH] add error checking to file reads
* Paul's Win32 service errors
* Dean's [PATCH] PR#1211: race condition in unblock_alarms
* Martin's [PATCH] Minor cleanups for portability
* Configure and "!: not found " error fixed
* Paul Eggert's [PATCH] mod_usertrack/1342: year-3000 bug
* Rasmus' [PATCH] Re: Module install steps
* Martin's [PATCH] update protocol/1329: wrong Content-length
* Brian Havard's [PATCH] OS2 fixes
* Dean and Roy's PTHREADS fixes
* Marc's [PATCH] PR#1352: mod_cgi logging broken
* Marc's [PATCH] FCNTL_SERIALIZED_ACCEPT on HPUX (PR#872)
* Brian Havard's [PATCH] PrintPathOS2 fixes
* Dean's [PATCH] Re: config/1368: Syntax error in src/Configure file
* Andrey A. Chernov's [PATCH] Re: os-freebsd/1311: fnmatch.c
* Dean's [PATCH] RAISE_SIGSTOP
* Jim's [PATCH] Make FLOCK mutex work PR#1056 (Take II)
* Marius Groeger <ma...@sysgo.de> [PORT] LynxOS 2.5
* Dean's [PATCH] PR#1248, 1328: mod_access problems (byte ordering)
* Lars' [PATCH] PR#870 preserve handler value on ProxyPass'ed requests
* Dean's [PATCH] Re: os-sunos/1294: Spurious mmap error messages? (fwd)
* Ken's new [PATCH] for mod_autoindex: no-sorting
* Ken's [PATCH] Year-2000 for mod_include default timefmt
* Martin's [PATCH] Show compile time switches for server
* Martin's [PATCH] More checks for command context in http_core
* Martin's [PATCH] Minor off-by-one error
* Dean's [PATCH] PR#1181: diagnose low descriptor situations
* Ralf's [PATCH] mod_rewrite ``Break Its Heart''
* Ken's [PATCH] ReadmeName and !FancyIndexing (PR#1373)
* Dean's [PATCH] pchild pool creation
* Ben's [PATCH] Apache Win32 v1.3b2 wouldn't serve >16k
* Ben Hyde's [PATCH] performance tweak.
* buildmark.c adjustment (use __DATE__ and __TIME__)
* SUNOS_LIB_PROTOTYPES/SunOS Build fixes protoypes.
* Ken's [PATCH] for AllowOverrides-not-in-VirtualHost
* Lars' [Patch] provide useful error entry if AuthType is missing
* Dean's [PATCH] clean up check_cmd_context
* Ken's [PATCH] start on libap (fixes PR#905 et alia)
* Marc's [PATCH] typecast in mod_mime_magic
Available:
* Paul's [PATCH] NT: spaces in argv[]
<Pi...@ecstasy.localnet>
Status: Paul +1, Ben +1
* Marc's [PATCH] fix strtoul
<Pi...@alive.znep.com>
Status: Marc +1, Jim +1, Martin +1, Dean +1
* Roy's [PATCH] protocol/1399: failing to read body
<97...@paris.ics.uci.edu>
Status: Roy +1,
* Paul Eggert's [PATCH] suexec/1343: year-2000 bug in suexec log
<Pi...@twinlark.arctic.org>
Status: Martin +1, Marc +1, Dean (Wants something more common)
* Martin's [PATCH] "Signing" server generated pages
<19...@deejai.mch.sni.de>
Status: Martin +1,
* Marc's [PATCH] define to allow passing of Authorization header
<Pi...@alive.znep.com>
Status: Marc +1, Dean +1 (with note), Paul +1
* Dean's protocol/1195: Bug in Authentication header (fwd)
<Pi...@twinlark.arctic.org>
Status: Conceptual: Dean +1, Paul +1, Martin +1, Ken +1
* Dean's [PRE-PATCH] expanding ap_snprintf()
<Pi...@twinlark.arctic.org>
Status: Dean +1, Ben +1, Jim 0, Martin 0, Brian +1(?)
See <Pi...@twinlark.arctic.org>
for a more up-to-date idea (int vformatter) that has a
vote of +1 from Dean, Ben, Martin, Paul and Jim for concept
* Martin's [PATCH] [FEATURE] Clickable Path Components in ftp dir header
<19...@deejai.mch.sni.de>
<19...@deejai.mch.sni.de>
Status: Martin +1, Ken +1
In progress:
* Martin Kraemer's [PATCH] Parsing URI into its components
This has "evolved" into a new module: util_uri. Martin
will post when it's at a state where he's happy with it.
Needs patch:
Closed issues:
* Removal of inetd mode
Ken says he'll try to maintain it, since there are
people/places who need it
Open issues:
* Ken suggests that new check_cmd_context() and related defines
should be non-static and in util_* so modules can use 'em. (He
didn't notice this flaw during the review.)
* Sameer's mod_so implemetation
See <19...@gabber.c2.net>
Issues: Underscores: Should I try prepending, appending, and
ignoring? -> Alexei says look at Java
Location? os/unix ??
* 206 vs. 200 issue on Content-Length
See <Pi...@valis.worldgate.com>
* Conditional logging and aplogger
Dean broke piped logging, -DNO_OTHER_CHILD to disable the reliable
stuff if you depend on piped logging. I figure nobody must, since
we haven't seen a bug report and it'd show up easily under moderate
load. Brian thinks Cond. logging is taking us in an opposite
direction from where we should be going.
* Brian's ``strange performance delays''
<3....@hyperreal.org>
mod_proxy and garbage collection?
see: <19...@deejai.mch.sni.de>
* Marc's socket options like source routing (kill them?)
Marc, Dean, Martin say Yes
* NameVirtualHost
Causes mucho confusion... Dean is ready to ``admit defeat''
but should we let him? Lars suggests making NameVirtualHost
a required directive or at least an error message, since
it will snag many who go from 1.2 to 1.3.
* DDA's adding a tray application to the Windoze version for ease of
status/management.
<01...@caravan.individual.com>
<01...@caravan.individual.com>
Status: Ken +1, Sameer +1, Martin +1, Ben +1 (as long as
we get a single executable)
Paul: No like Win95 specific stuff
* The PR#918 patch above appears to have been bogus, should be fixed or
reversed. It has been reversed for now.
* Martin's SEGV, and Alexei's GP Fault under windows (probably the same)
* Marc's [BUG] include virtual and SCRIPT_NAME w/path_info
<Pi...@alive.znep.com>
--
====================================================================
Jim Jagielski | jaguNET Access Services
jim@jaguNET.com | http://www.jaguNET.com/
"Look at me! I'm wearing a cardboard belt!"
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Ben Laurie <be...@algroup.co.uk>.
Jim Jagielski wrote:
>
> Very, very close... At present, it looks like the
> only thing is the ``spaces in NT argv'' problem.
> Paul has a suggested patch, which Ben agrees with,
> but Ben points out that a more general solution
> is needed. He will try to work one out. We'll hold
> off on 1.3b3 until Ben decides.
OK, I've more or less done it, and it seems to work. Just a tad more
tidying up and I'll commit.
Cheers,
Ben.
--
Ben Laurie |Phone: +44 (181) 735 0686|Apache Group member
Freelance Consultant |Fax: +44 (181) 735 0689|http://www.apache.org
and Technical Director|Email: ben@algroup.co.uk |Apache-SSL author
A.L. Digital Ltd, |http://www.algroup.co.uk/Apache-SSL
London, England. |"Apache: TDG" http://www.ora.com/catalog/apache
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Ben Laurie <be...@algroup.co.uk>.
Jim Jagielski wrote:
>
> Very, very close... At present, it looks like the
> only thing is the ``spaces in NT argv'' problem.
> Paul has a suggested patch, which Ben agrees with,
> but Ben points out that a more general solution
> is needed. He will try to work one out. We'll hold
> off on 1.3b3 until Ben decides.
BTW, there's also an assert I'm going to blow away.
And there was some debate as to whether os_canonical_filename should
lowercase the parts of the file that _don't_ correspond to actual files.
I think I agree that it shouldn't (and it currently does) but I don't
intend to hold up 1.3b3 for it - can we add it to the issues list for
1.3b4, please?
Cheers,
Ben.
--
Ben Laurie |Phone: +44 (181) 735 0686|Apache Group member
Freelance Consultant |Fax: +44 (181) 735 0689|http://www.apache.org
and Technical Director|Email: ben@algroup.co.uk |Apache-SSL author
A.L. Digital Ltd, |http://www.algroup.co.uk/Apache-SSL
London, England. |"Apache: TDG" http://www.ora.com/catalog/apache
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Ben Laurie <be...@algroup.co.uk>.
Martin Kraemer wrote:
> > 2. Because there is only one, there can be some pretty long delays when
> > trying to serve multiple connections - you have to wait for the first
> > one's keepalive to time out...
> But it only talked to one client process; shouldn't these two have dealt
> with it correctly?
If you are using Netscape with out-of-the-box settings, then it tries to
start 4 (or 2 for later versions?) simultaneous connections. Dunno for
MSIE.
Some combination of ctl, alt, shift and T shows you the current
connections. I forget excatly what.
Cheers,
Ben.
--
Ben Laurie |Phone: +44 (181) 735 0686|Apache Group member
Freelance Consultant |Fax: +44 (181) 735 0689|http://www.apache.org
and Technical Director|Email: ben@algroup.co.uk |Apache-SSL author
A.L. Digital Ltd, |http://www.algroup.co.uk/Apache-SSL
London, England. |"Apache: TDG" http://www.ora.com/catalog/apache
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Martin Kraemer <Ma...@mch.sni.de>.
On Sun, Nov 16, 1997 at 10:28:14PM +0100, Martin Kraemer wrote:
> That's what I thought as well. Seems I had inadvertently left some mod_fastcgi
> stuff is there which misbehaved VERY badly (it was 1.1 or 1.2 code).
> Ok, I removed that stuff.
Yup, that was the reason. Okay, no reasons from my side why 1.3b3 should
not be rolled.
Martin
--
| S I E M E N S | <Ma...@mch.sni.de> | Siemens Nixdorf
| ------------- | Voice: +49-89-636-46021 | Informationssysteme AG
| N I X D O R F | FAX: +49-89-636-44994 | 81730 Munich, Germany
~~~~~~~~~~~~~~~~My opinions only, of course; pgp key available on request
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Martin Kraemer <Ma...@mch.sni.de>.
On Sun, Nov 16, 1997 at 09:00:11PM +0000, Ben Laurie wrote:
>
> Eh? Two things:
>
> 1. With -X there should be no more than one httpd.
That's what I thought as well. Seems I had inadvertently left some mod_fastcgi
stuff is there which misbehaved VERY badly (it was 1.1 or 1.2 code).
Ok, I removed that stuff.
> 2. Because there is only one, there can be some pretty long delays when
> trying to serve multiple connections - you have to wait for the first
> one's keepalive to time out...
But it only talked to one client process; shouldn't these two have dealt
with it correctly?
I'm just recompiling the latest version (without fastcgi) to see it things
changed.
Martin
--
| S I E M E N S | <Ma...@mch.sni.de> | Siemens Nixdorf
| ------------- | Voice: +49-89-636-46021 | Informationssysteme AG
| N I X D O R F | FAX: +49-89-636-44994 | 81730 Munich, Germany
~~~~~~~~~~~~~~~~My opinions only, of course; pgp key available on request
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Ben Laurie <be...@algroup.co.uk>.
Martin Kraemer wrote:
>
> On Sun, Nov 16, 1997 at 10:17:05AM -0500, Jim Jagielski wrote:
> > Very, very close... At present, it looks like the
> > only thing is the ``spaces in NT argv'' problem.
> > Paul has a suggested patch, which Ben agrees with,
> > but Ben points out that a more general solution
> > is needed. He will try to work one out. We'll hold
> > off on 1.3b3 until Ben decides.
> > ==================================================
>
> One thing I noticed this weekend -- didn't try to nail it down yet though --
> When the server is started with the "-X" switch, it sometimes locks up
> after serving a cpuople of documents. I mean, it still listens on ots port,
> but it doesn't read (and answer) any requests any more. And it does this
> before its max number of requests is reached.
>
> I observed this behavior both on Linux and on SVR4.
> Can anyone else reproduce this?
>
> At the moment of the lockup, I see three httpd's, one is blocked in
> read(4,..), when I try to attach strace to the others nothing is
> printed -- (not in a sys call?!)
Eh? Two things:
1. With -X there should be no more than one httpd.
2. Because there is only one, there can be some pretty long delays when
trying to serve multiple connections - you have to wait for the first
one's keepalive to time out...
Cheers,
Ben.
--
Ben Laurie |Phone: +44 (181) 735 0686|Apache Group member
Freelance Consultant |Fax: +44 (181) 735 0689|http://www.apache.org
and Technical Director|Email: ben@algroup.co.uk |Apache-SSL author
A.L. Digital Ltd, |http://www.algroup.co.uk/Apache-SSL
London, England. |"Apache: TDG" http://www.ora.com/catalog/apache
-X switch (was Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997)
Posted by Dean Gaudet <dg...@arctic.org>.
Known problem ... -X and inetd mode don't have working timeouts since my
OPTIMIZE_TIMEOUT code went in. To fix them requires a small tweak to
set_callback_and_timeout() to use alarm() instead of the scoreboard when
in either mode.
Dean
On Sun, 16 Nov 1997, Martin Kraemer wrote:
> One thing I noticed this weekend -- didn't try to nail it down yet though --
> When the server is started with the "-X" switch, it sometimes locks up
> after serving a cpuople of documents. I mean, it still listens on ots port,
> but it doesn't read (and answer) any requests any more. And it does this
> before its max number of requests is reached.
>
> I observed this behavior both on Linux and on SVR4.
> Can anyone else reproduce this?
>
> At the moment of the lockup, I see three httpd's, one is blocked in
> read(4,..), when I try to attach strace to the others nothing is
> printed -- (not in a sys call?!)
>
> Martin
> --
> | S I E M E N S | <Ma...@mch.sni.de> | Siemens Nixdorf
> | ------------- | Voice: +49-89-636-46021 | Informationssysteme AG
> | N I X D O R F | FAX: +49-89-636-44994 | 81730 Munich, Germany
> ~~~~~~~~~~~~~~~~My opinions only, of course; pgp key available on request
>
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Martin Kraemer <Ma...@mch.sni.de>.
On Sun, Nov 16, 1997 at 10:17:05AM -0500, Jim Jagielski wrote:
> Very, very close... At present, it looks like the
> only thing is the ``spaces in NT argv'' problem.
> Paul has a suggested patch, which Ben agrees with,
> but Ben points out that a more general solution
> is needed. He will try to work one out. We'll hold
> off on 1.3b3 until Ben decides.
> ==================================================
One thing I noticed this weekend -- didn't try to nail it down yet though --
When the server is started with the "-X" switch, it sometimes locks up
after serving a cpuople of documents. I mean, it still listens on ots port,
but it doesn't read (and answer) any requests any more. And it does this
before its max number of requests is reached.
I observed this behavior both on Linux and on SVR4.
Can anyone else reproduce this?
At the moment of the lockup, I see three httpd's, one is blocked in
read(4,..), when I try to attach strace to the others nothing is
printed -- (not in a sys call?!)
Martin
--
| S I E M E N S | <Ma...@mch.sni.de> | Siemens Nixdorf
| ------------- | Voice: +49-89-636-46021 | Informationssysteme AG
| N I X D O R F | FAX: +49-89-636-44994 | 81730 Munich, Germany
~~~~~~~~~~~~~~~~My opinions only, of course; pgp key available on request
Re: [STATUS] 1.3b3-dev Sun Nov 16 10:07:28 EST 1997
Posted by Marc Slemko <ma...@worldgate.com>.
On Sun, 16 Nov 1997, Jim Jagielski wrote:
>
> Plan:
>
> * At present, we are close to releasing 1.3b3. We are waiting
> for a resolution of the "blank in filenames" problem in Win32
> before we tag and freeze 1.3b3. There are some patches that
> have enough votes to be included in 1.3b3 if commited soon.
> When the above mention "problem" is fixed, the 1.3b3 code
> will be CVS tagged, but a tarball won't be rolled. This is
> so we can roll the UNIX stuff and the Win32 stuff at the same
> time, at which point we can formally announce. Ken wants to
> do the announcement.
Would it perhaps be an idea to roll the Unix tarball before and
let people test it more before release? Saves time later.
Also note my "don't include netdb.h in mod_unique_id.c" patch.