Hi Igor,
attached patch fixes ./configure script for less error-forgiving compilers
(like LLVM/clang) when they are used with -Werror option.
auto/types/sizeof:
- printf() is defined in missing <stdio.h> header,
- sizeof() returns size_t type, which has different signedness on different
operating systems (hence %z).
auto/unix:
- setproctitle() is defined in <stdlib.h> on both: NetBSD and OpenBSD.
Best regards,
Piotr Sikora < piotr.sikora(a)frickle.com >
Hi all, I would like to contribute a patch to provide the cached file
age as a variable $proxy_cache_age , which can then be used together
with add_header Age $proxy_cache_age.
The age calculation is not RFC2616 compliant (13.2.3) as it doesn't take
into consideration the cached origin Age header and Date. Is it feasible
to extract this from the cached headers?
--
regards,
wK
XSLT parameter parsing currently allows paramters to be passed one per
line or multiple per line separated by a colon (:).
http://wiki.nginx.org/NginxHttpXsltModule#xslt_stylesheet
location / {
xslt_stylesheet /site/xslt/one.xslt
$arg_xslt_params
param1='$value1': param2=value2
param3=value3;
}
This has the unfortunate effect of preventing xpath parameters being
passed as url parameters when they contain a colon (fairly common in
xpath as a namespace or axis specififer). Consider the following
configuration:
location / {
xslt_stylesheet filter.xsl
xpath=$arg_xpath;
}
I would like to be able to call call the url
/doc.xml?xpath=ancestor-or-self::entry, but the current parameter
parsing code interprets this as multiple invalid parameters
``xpath=ancestor-or-self``, ````, ``entry``. As far as I can tell
there is no way to substitute ":" for "%3A" in a variable.
The attached patch removes support for specifying multiple parameters
in a single line. You may still supply multiple parameters to an xslt
by supplying one per line. (I've split the patch into two changesets
for readability.)
As an alternative it may be possible to preserve the current behaviour
by splitting parameters before the call to ngx_http_complex_value
which performs the variable substitution, however I was unable to work
out how to do this myself.
Laurence
I want to write mod_layout the right way.
layout have to insert "header" after the body element, or at the top it
there is no body element, but
there should be no header when there is no body and there is </frameset> on
the page.
I modified sub_filter_module:
http://rkd.republika.pl/NGINX/ngx_http_sub2_filter_module.c
Config options are similar to sub_filter.
I'm using:
sub2_filter <body 'HEADER<br/>';
sub2_filter_once on;
sub2_frameset </frameset>;
Sometimes it works, but when I set output_buffers to 1 1; it stopped working
with html like this:
<frameset cols="50%,50%">
<frame name ="a" src="a.html"/>
<frame name="b" src="b.html"/>
</framset>
How to make it work, especially how to achieve inserting something before
whole document, when there is no <body> in the document?
Does somebody know if it's possible to specify nginx to get special urls or files from other nginx? I want to bring up nginx on separate port to have possibility to test it without custom modules.
Thanks.
Hi, I looked at the 0.8.49 to .50 diff today and came across the
function "ngx_http_set_comlex_value_slot()" , looking at the function
body it looks like it should be spelled "ngx_http_set_complex_value_slot()".
--
regards,
wK
Hi guys,
I hope nobody minds me asking for help in this way.
I'm asking for help from programmers to help set up a unique open-source
project that could benefit a great many people.
The Tagmata Project's three aims are :
(1) to create a completely open-source website
(2) to use the site to generate money
(3) to use the profits to help alleviate poverty
Full details of the whole project can be found at http://www.tagmata.info.
The platform for the open-source website has not yet been built, and
that is what I'm asking for help with.
There are two things in particular that seem most relevant to Nginx
developers / users :
(1) Nginx - MongoDB module
The Tagmata database will use MongoDB, with an Nginx front-end. We're
looking to develop an Nginx module that connects to MongoDB, to allow
queries through Nginx to MongoDB (in a similar vein to the recent
ngx_drizzle, ngx_postgres etc modules that have been developed recently
by Agentzh, Piotr Sikora et al.). This might be able to reuse some code
from the ngx_gridfs module, but I'm not sure how stable it is.
Another part of this module will be translating our own database query
language (Tagmata Query Language) to MongoDB queries - which should be
pretty easy, as the syntax is very similar.
The code for this will be hosted on GitHub, and will be designed in such
a way that it will be reusable for other projects that wish to connect
to MongoDB that don't want to use Tagmata Query Language.
Once this particular module is written and stable, I'll post here to let
Nginx users know.
(2) Online development tools
In order to facilitate adding/modifying of code on the open-source
website, we're developing our own tools to make this easy. These will be
written in our own scripting language (a cross between PHP and
Javascript/Ruby - see http://www.tagmata.info/development/info/TPL for
details).
The code for these will be made available on the open-source website
(tagmata.com) when the base platform has been developed.
I can do all of this code myself, but I'm hoping that some generous
programmers would be willing to help out to make the task easier. Some
other programmers have volunteered to do some of the other tasks that
need doing, but we could do with some more help to speed up the
development process.
If you'd like to know more about the project, please visit
http://www.tagmata.info.
If you'd like to know more about the programming tasks, please visit
http://www.tagmata.info/development/programming.
If you'd like to get involved, please contact me by email
(dev(a)tagmata.com) or IRC (Freenode #tagmata).
Thanks in advance,
Marcus Clyne.
Igor and Maxim,
all stack traces I reported before were caused by glibc bug in CentOS that corrupted memory. It was the reason on nginx faul in random place. I's sorry for taking your time and thanks for loooking into this issue.
In CentOs MALLOC_CHECK_ is enabled by default. And causes memory corruption for applications.
More info is here:
http://www.pubbs.net/200911/kde/23500-deactivate-malloccheck-on-x8664.html
Patch to fix it is here:
http://sourceware.org/bugzilla/show_bug.cgi?id=10282
So it's possible to fix glibc or export MALLOC_CHECK_=3 to avoid random crashes (which unfirtunatelly began happen after enabling cache).