Lighttpd

2007-02-02

16:01 Revision 1587 (svn): reordered includes to make glib + our (buffer|array) happy

jan

15:41 Revision 1586 (svn): glib 2.4.0 is good enough, perhaps even older ones

jan

15:35 Revision 1585 (svn): move compile fixes for glib-2.4.0

jan

15:28 Revision 1584 (svn): let's try to compile on glib2 2.4.0

jan

15:27 Revision 1583 (svn): include filename into the error-msg

jan

15:17 Revision 1582 (svn): make gcc happy about the 'array' as param-name and 'array' as type-name

jan

15:17 Revision 1581 (svn): removed unused TRUE/FALSE defines which clash with glib

jan

05:30 Feature #1004 (Wontfix): Consider using Lua as the configuration file language.

Consider using Lua as the configuration file language.
Lighttpd already uses and supports Lua for some things (e.g... Anonymous

2007-02-01

19:17 Feature #322: FastCGI Authorizer support for Variable-name variable passing

Replying to [[lighttpd:comment:5|EDevil]]:
> I think that with this patch we still have to opt between serving stati... maherb

18:55 Revision 1580 (svn): added url.redirect-code

jan

13:51 Revision 1579 (svn): honour srv->is_shutdown and leave the thread

jan

13:45 Revision 1578 (svn): raised the chunk-size to get better performance for large file-transfers

jan

12:58 Revision 1577 (svn): possibly fixed a fdevent which is running away

jan

12:57 Revision 1576 (svn): handle out-of-fd condition nicely

jan

12:41 Revision 1575 (svn): use sync-read for small files

jan

12:33 Revision 1574 (svn): added missing cur_fds-- to lift the 'out-of-fds' condition again

jan

12:24 Revision 1573 (svn): improved error-msg to include more useful debug-info

jan

11:38 Revision 1572 (svn): join all the threads we started and cleanup after them

jan

11:17 Revision 1571 (svn): moved the completion-handlers into the network-backends

* server.c, network_posix_aio.c, network_linux_aio.c
- added counters for sync/async reads
- moved the completion... jan

10:48 Bug #499: FastCGI stderr output doesn't go to error log

Yes, a properly written FastCGI app will never produce output to stderr. However, the point of sending stderr output ... Anonymous

08:39 Revision 1570 (svn): fixes cmake build for FreeBSD systems, removed "-lrt" and replaced "-ldl" with "-lc".

someone with a FreeBSD system will need to test this change, it still works on Linux. jakabosky

02:32 Revision 1569 (svn): moved the io_getevents() into the network-backend

- improved performances as the iocbs are released earlier jan

02:31 Revision 1568 (svn): declared functions as static

jan

02:06 Bug #541: Pdf problem with Internet Explorer documentation/default config

I've also received reports of this problem with Lighttpd 1.4.13. Apparently also works fine for FF.
Funny thing is... Anonymous

01:30 Revision 1567 (svn): for small files read synchronously

jan

01:21 Revision 1566 (svn): fixed bug with run-time undefined symbols for buffer_append_off_t & deflate

jakabosky

01:12 Revision 1565 (svn): added a extra API call for the async-stat to make old modules behave as

before jan

01:06 Revision 1564 (svn): honour async-stat() call

jan

00:58 Revision 1563 (svn): removed debug message

jan

00:44 Revision 1562 (svn): improved the pkg-config checks for glib-2.0

jan

00:17 Revision 1561 (svn): big threading cleanup, glib-2.0 + gthread-2.0 is now a requirement

* configure.in, src/CMakeLists.txt
- removed FAM support
- added checks for glib-2.0
* src/Makefile.am
- added ... jan

2007-01-31

18:03 Feature #1003: Feature request: X-Subrequest-*

what would be the usecase? except generating useless cpu load?
darix

16:30 Feature #1003 (Wontfix): Feature request: X-Subrequest-*

Apache + mod_php include a special function called virtual() that lets you start a new subrequest inside apache and n... Anonymous

15:55 Bug #1002 (Fixed): 304 (not modified) must have no Content-Length

According to RFC 2616 (section 4.3), these responses MUST NOT include a message body: 1xx, 204, 304. This means they ... Anonymous

11:19 Bug #998: fastcgi prints an additional empty line for each packet of type FCGI_STDERR

the problem with PHP is a bug in PHP not in lighttpd or fastcgi as error_log is printing a an extra "CRLF" at the end... carenas

02:12 Bug #476 (Fixed): HTTP/1.0 Request Crash lighttpd when using mod_proxy

fixed in r901 1.4.x branch and included in releases since atleast 1.4.11
jakabosky

00:42 Bug #397 (Fixed): lighttpd 1.4.8 ssl problem

this just means someone tries to use non ssl connections on your ssl port. no bug here.
please dont reopen the bug
darix

2007-01-30

22:18 Revision 1560 (svn): - fixing syntax of the targetlibs for mod_magnet

darix

20:33 Revision 1559 (svn): Fixed bug with X-Rewrite-* headers

jakabosky

15:35 Revision 1558 (svn): added inotify support

jan

15:34 Revision 1557 (svn): play nice if we run out of fds

jan

13:10 Bug #397 (Need Feedback): lighttpd 1.4.8 ssl problem

I get a lot of these messages in my error logs (actually these are the only errors I get), running lighttpd 1.4.13 bu... Anonymous

12:44 Revision 1556 (svn): added EPIPE and a better error-msg for debugging

jan

12:42 Revision 1555 (svn): wake up a connection which is waiting on network-write() when it ran out

of fds jan

12:42 Revision 1554 (svn): report a out-of-fd nicely to the upper layers

jan

12:41 Revision 1553 (svn): if a module sets Content-Length: itself we would append our own header a

second time creating a duplicate header jan

12:05 Revision 1552 (svn): ported mod_magnet to the 1.5.0 API

jan

11:17 Revision 1551 (svn): added mod_magnet to the static-build

jan

11:17 Revision 1550 (svn): use internal md5 'const' where it is possible

jan

11:16 Revision 1549 (svn): liblua wants libm too

jan

11:15 Revision 1548 (svn): added trailing newline at end-of-file to make the c-compiler happy

jan

11:01 Revision 1547 (svn): added mod_magnet to the cmake compilation

jan

10:50 Revision 1546 (svn): copied mod_magnet from the 1.4.x branch

jan

04:16 Bug #993 (Fixed): server.upload-dirs is ignored

fix in r1545.
jakabosky

04:16 Revision 1545 (svn): fixed bug with "server.upload-dirs" (fixed #993)

tempfile chunks are allocated from con->recv no con->recv_raw jakabosky

03:08 Bug #986 (Fixed): Response breaking in mod_deflate

fixed in r1544.
The problem was with deflate.work-block-size = 64 = (64 * 1024) = 65536.
jakabosky

03:00 Revision 1544 (svn): Fixed major bug with compressing chunks with offset > 0 or when deflate.work-block-size > 0 is used

(fixed #986) jakabosky

01:48 Bug #653: mod_cgi doesn't work with server.bind set to unix domain socket

I'm having the same problem (with the same result: a zero-length "copy" of the script is downloaded by the browser) w... Anonymous

01:32 Bug #996 (Fixed): mod_deflate should allow limiting of compression algorithm from the configuration file

fixed in r1543.
new option:
deflate.allowed_encodings = ( "bzip2", "gzip", "deflate" )
or
deflate.allowed_en... jakabosky

01:28 Revision 1543 (svn): replaced deflate.bzip2 option with deflate.allowed_encodings.

allowed_encodings is an array with the compression types to enable. jakabosky

00:30 Bug #998: fastcgi prints an additional empty line for each packet of type FCGI_STDERR

this patch however doesn't solve problem with PHP-FCGI error logs:... Anonymous

2007-01-29

22:12 Bug #996: mod_deflate should allow limiting of compression algorithm from the configuration file

added mod_deflate to components list.
jakabosky

19:23 Bug #1001: Proxy Secure connections Error : (connections.c.538) connection closed: write failed on fd 8

Replying to [[lighttpd:comment:2|darix]]:
> the http proxy module doesnt speak https. your proxy backend needs to sp... kdlarg

17:46 Bug #1001: Proxy Secure connections Error : (connections.c.538) connection closed: write failed on fd 8

the http proxy module doesnt speak https. your proxy backend needs to speak plain http.
does the port 443 in your pr... darix

17:20 Bug #1001: Proxy Secure connections Error : (connections.c.538) connection closed: write failed on fd 8

Sorry let me add more info from the error log:
2007-01-29 12:00:51: (request.c.293) fd: 8 request-len: 438
GET / ... kdlarg

17:10 Bug #1001 (Invalid): Proxy Secure connections Error : (connections.c.538) connection closed: write failed on fd 8

Hello,
We have a few lighttpd servers running proxies and images. We love it.
However we have come across an is... kdlarg

18:11 Bug #988 (Fixed): spawn-fcgi does not really redirect stderr and stdout

fixed in r1542
jan

18:11 Revision 1542 (svn): swapped params in dup2() (fixed #988)

jan

18:10 Revision 1541 (svn): added option to enable/disable libfcgi support for fcgi-stat-accel

jan

07:18 Bug #1000 (Fixed): bad accesslog.format config can cause lighttpd-1.4.11 to segfault on startup

using `accesslog.format = "%{%}a"` in lighttpd.conf causes lighttpd-1.4.11 to segfault on startup.
-- tkooda Anonymous

2007-01-28

19:11 Revision 1540 (svn): - applied missing chunk from the static balancer patch

darix

16:08 Bug #999 (Obsolete): mod_flv_streaming

I've downloaded lighttpd with darwinports specifically for streaming flv files.
The version I have is 1.4.13. How... Anonymous

14:41 Bug #998: fastcgi prints an additional empty line for each packet of type FCGI_STDERR

a fastcgi program using the fcgi_stdio emulation is shown below:... carenas

10:00 Bug #998: fastcgi prints an additional empty line for each packet of type FCGI_STDERR

in lighttpd 1.5 the implementation used for mod_proxy_backend_fastcgi doesn't show this problem as shown in :
<pr... carenas

07:25 Bug #998 (Fixed): fastcgi prints an additional empty line for each packet of type FCGI_STDERR

every time a fastcgi script writes an error using (for example for the C API) :
FCGX_FPrintF(err, "This is an er... Anonymous

14:34 Revision 1539 (svn): added static balancer ( submitted by uwe.voelker@gmx.de )

jan

12:48 Bug #499: FastCGI stderr output doesn't go to error log

just to clarify:
sending errors directly to the log violates the fastcgi specification as it specifically says tha... carenas

08:18 Bug #499: FastCGI stderr output doesn't go to error log

this is still present on the last release and the semantics of a fix are complicated by the fact that the fastcgi pro... Anonymous

12:47 Revision 1538 (svn): fixed checks for libfcgi and the different locations of the

fastcgi-headers jan

08:04 Bug #729: Incorrect PATH_INFO when FastCGI is serving "/"

FWIW, the above patch fixes this bug for mod_scgi as well (in my case, tested with DrProject).
-- shot Anonymous

06:32 Feature #507: optional consistent ETag header base on file content instead of only inode

This would be helpful in my setup also.
I lost this when I move from apache to lighttpd.
(Noone other than me ha... Anonymous

05:07 Bug #997 (Fixed): lighttpd -V Missing linux-aio-sendfile

I have linux-aio-sendfile compilled in (i think) yet in lighttpd -V only lists sendfile
Adding
server.network-bac... Anonymous

00:13 Revision 1537 (svn): moved the reset of the content-queue right before we write the content

to the network as all filters might want to know if there was content
and set the Content-Length accordingly. jan

2007-01-27

21:20 Revision 1536 (svn): announce that we have no content to send in case of a parse-error in the

HTTP-Request jan

21:17 Revision 1535 (svn): build fcgi-stat-accel in case we have libfcgi

jan

18:41 Revision 1534 (svn): fixed out-of-fd detection

instead of testing activly if we will run out of fds soon we just let it
fail once and handle it nicely by disabling ... jan

18:39 Revision 1533 (svn): added fcgi-stat-accel and spawn-fcgi to the build

jan

16:51 Revision 1532 (svn): added fcgi-stat-accel to the distribution

jan

16:21 Bug #996 (Fixed): mod_deflate should allow limiting of compression algorithm from the configuration file

A configuration option should be added to select which of gzip/deflate/bzip2 are allowed in a particular context.
... jrabbit

16:07 Revision 1531 (svn): reverted a small part of [1522] as it breaks the normal operation of the

HTTP/1.0 proxying if Connection: close is set jan

15:46 Feature #965: New conditional: Physical path

This was implemented (for 1.5) in r1528.
hoffie

14:49 Feature #761: client certificate support for mod_auth ldap backend

the fix only applies for 1.5.x.
there are no plans to backport it to 1.4.x
darix

14:31 Feature #761 (Fixed): client certificate support for mod_auth ldap backend

applied in r1530
jan

14:31 Revision 1530 (svn): added support for client-certs to auth against the ldap-server (taken

from #761) jan

14:23 Bug #990 (Fixed): proxy-core.balancer should not be mandatory

fixed in r1529, we use round-robin as default. If there is only one backend available it is just ignored.
jan

14:21 Revision 1529 (svn): don't require a value for proxy-core.balancer, assume round-robin

jan

14:14 Revision 1528 (svn): added $PHYSICAL["path"] and $HTTP["request-method"]

* configparser.y
- added the new options and added alternative names for user-agent,
remote-ip and query-string jan

13:28 Revision 1527 (svn): in 1.4.x it is X-LIGHTTPD-send-file, be compatible with that

jan

12:59 Revision 1526 (svn): removed strlen() from the array_get_element()

* *.c
- use CONST_STR_LEN() as most strlen()s are already known
- use BUF_STR_LEN() for the buffer jan

06:25 Bug #971 (Fixed): spawn-fcgi ignores php command line arguments like "-c /var/www/userxyz/php/php.ini"

darix

04:18 Bug #971: spawn-fcgi ignores php command line arguments like "-c /var/www/userxyz/php/php.ini"

Replying to [[lighttpd:comment:6|ice@qgabi.de]]:
> ...
> PHP 5.2 is not available for Debian and Ubuntu.
> ...
... Anonymous