Re: [Libevent-users] evhttp server on unix socket

Le 01/12/2010 07:47, Basile Starynkevitch a écrit : > I don't know fur sure, and I believe it should require some patch. > > However, I am interested to understand why you are asking that. > > In particular, at least under Linux, I know unfortunately no Web > browser (or other HTTP client) able to

Re: [Libevent-users] evhttp server on unix socket

Le Wed, Dec 01, 2010 at 07:47:36AM +0100, Basile Starynkevitch ecrivait : > However, I am interested to understand why you are asking that. > > In particular, at least under Linux, I know unfortunately no Web > browser (or other HTTP client) able to send HTTP requests on an Unix > socket. REST

Re: [Libevent-users] evhttp server on unix socket

On Wed, 01 Dec 2010 07:28:25 +0100 Christophe Zbierski wrote: > Hi, is it possible to bind an HTTP server on unix socket instead of > address and port (int evhttp_bind_socket > (struct > > evhttp *ht

[Libevent-users] evhttp server on unix socket

Hi, is it possible to bind an HTTP server on unix socket instead of address and port (int evhttp_bind_socket (struct evhttp *http, const char *address, u_short port) ). Thanks Christophe

Re: [Libevent-users] can libevent-devel for x86_64 and i686 coexist on a x86_64 machine?

It's generated by autoconf *as part of the build process for Libevent*. If you build a new libevent with --prefix=/usr and do "sudo make install" on it, then you will overwrite /usr/include/event2/event-config.h. But if you're using a packaging system, then you shouldn't ever do that. Ah, r

Re: [Libevent-users] can libevent-devel for x86_64 and i686 coexist on a x86_64 machine?

On Tue, Nov 30, 2010 at 6:10 PM, Mr Dash Four wrote: > >> That's plausible; packagers should probably indeed do something like >> this; I hope that the package maintainer accepts it or something likes >> it. >> > > Tested it and it works to absolute perfection! I've done the changes exactly > as I

Re: [Libevent-users] can libevent-devel for x86_64 and i686 coexist on a x86_64 machine?

That's plausible; packagers should probably indeed do something like this; I hope that the package maintainer accepts it or something likes it. Tested it and it works to absolute perfection! I've done the changes exactly as I listed them in my previous post and have also created a small 'he

[Libevent-users] If you have a patch or a bug, *please* put it on the bugtracker

Hi, all. I want to highlight a paragraph from the 2.0.9-rc announcement: > We're almost out of known bugs with the 2.0.x series. If you know > about a bug that 2.0.9-rc does not fix, please have a quick look at > the bug tracker to make sure that there is an entry there. If there > is not, then

Re: [Libevent-users] can libevent-devel for x86_64 and i686 coexist on a x86_64 machine?

On Tue, Nov 30, 2010 at 10:22 AM, Mr Dash Four wrote: > >>> I will file a bug with Fedora because this needs to be fixed - as things >>> stand now it won't work as event-config.h will get overwritten by >>> whatever >>> package was installed last which is plain wrong. Thanks for your input! >>> >>

Re: [Libevent-users] ANN: Libevent 2.0.9-rc is released

On Tue, Nov 30, 2010 at 9:41 AM, Kelly Brock wrote: > Great news but the iocp fix needs a minor fix: > > Line 618 & 619 need to be modified as follows: >                setsockopt(sock, SOL_SOCKET, SO_UPDATE_ACCEPT_CONTEXT, >                    (char *)&as->lev->fd, sizeof(&as->lev->fd)); > > This

Re: [Libevent-users] ANN: Libevent 2.0.9-rc is released

Thanks a lot On Tue, Nov 30, 2010 at 5:31 PM, Frank Denis wrote: > Le Tue, Nov 30, 2010 at 05:02:22PM +0200, Sherif Fanous ecrivait : > > The Changelog states the following > > > > Disable changelist for epoll by default because of Linux dup() bug; add > an > > option and/or an envvar to reenabl

Re: [Libevent-users] ANN: Libevent 2.0.9-rc is released

Le Tue, Nov 30, 2010 at 05:02:22PM +0200, Sherif Fanous ecrivait : > The Changelog states the following > > Disable changelist for epoll by default because of Linux dup() bug; add an > option and/or an envvar to reenable it for speed. > > Can an example be provided on how to enable it please H

Re: [Libevent-users] ANN: Libevent 2.0.9-rc is released

The Changelog states the following Disable changelist for epoll by default because of Linux dup() bug; add an option and/or an envvar to reenable it for speed. Can an example be provided on how to enable it please Thanks On Tue, Nov 30, 2010 at 4:46 PM, Ralph Castain wrote: > And I have a Win

Re: [Libevent-users] ANN: Libevent 2.0.9-rc is released

And I have a Windows and Solaris compile patch for you too :-) My bad - I'm behind in getting it to you. On Nov 30, 2010, at 7:41 AM, Kelly Brock wrote: > Great news but the iocp fix needs a minor fix: > > Line 618 & 619 need to be modified as follows: > setsockopt(sock, SOL_SOCK

RE: [Libevent-users] ANN: Libevent 2.0.9-rc is released

Great news but the iocp fix needs a minor fix: Line 618 & 619 need to be modified as follows: setsockopt(sock, SOL_SOCKET, SO_UPDATE_ACCEPT_CONTEXT, (char *)&as->lev->fd, sizeof(&as->lev->fd)); This is due to the line just obove setting as->s to invalid. KB >