VoIP Mailing List Archives

[trevor at concipient.net]

Has there been any progress getting FreeSWITCH to build on Ubuntu
Intrepid without downgrading libtool?

Thanks!

Sincerely,
Trevor Hammonds

_______________________________________________
Freeswitch-users mailing list
Freeswitch-users@lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

[anthony.minessale at g...]

We do not support ubuntu interpid, it has at least 3 known fatal issues not experienced by all but nonetheless enough to make us unwilling to support it.

It's "use at your own risk" or use the stable branch "hardy" for any support.



On Thu, Mar 26, 2009 at 4:45 AM, Trevor Hammonds <trevor@concipient.net (trevor@concipient.net)> wrote:

[dujinfang at gmail.com]

On Mar 26, 2009, at 10:59 PM, Anthony Minessale wrote:

[mike at jerris.com]

Another example of a fatal issue was the optimizer in gcc was breaking openzap code even with -O2.

Mike

On Mar 27, 2009, at 12:12 PM, dujinfang wrote:

[gkuri at ieee.org]

regarding gcc compiler optimizations, are they generally compatible with
FS or should they be removed or does the configure strip them out? just
curious, as I run Gentoo and use such optimizations as "-march=nocona
-O2 -pipe -fomit-frame-pointer"

not sure if they break things or I should be removing them before
compiling FS?

Gabe

Michael Jerris wrote:

[brian at freeswitch.org]

Usually if you don't know what they do... then you shouldn't use them!



/b

On Mar 27, 2009, at 1:19 PM, Gabriel Kuri wrote:

[gkuri at ieee.org]

I'm not asking what they do, I'm asking those more familiar with FS
whether the optimization flags are too aggressive for FS. What do you
guys (developers) normalize use, just your basic -march=i686 -pipe ?

Gabe


Brian West wrote:

[brian at freeswitch.org]

We usually don't specify anything extra!

/b

On Mar 27, 2009, at 1:56 PM, Gabriel Kuri wrote:

[anthony.minessale at g...]

1) There is an incompatibility on the fake ncurses wrapper that causes an instant seg fault unless you install the real ncurses.
2) The bleeding edge GCC builds an openzap binary that crashes instantly with no explanation in the core file from a minimal -O2 (that's just the one copmiler bug that we know about for sure, like cock roaches, see one, there are probably 1000)
3) They upgraded to libtool 2.0 which builds binaries that will not start. (easier said than done to upgrade ours too as we have to make sure we work on *every* plarform and the upgrade to make it work would break other operating systems we support)

Bottom line, it's not their fault or anything but the choice to use all brand new versions of everything under the sun is not a good idea for your server, it's great that we have bleeding edge stuff or we would not have anyone to test stuff, we have a similar group of people always running SVN trunk of the day.  But it's hard to stabalize code when both your code and the OS may be unstable at the same time.

There is a reason they call it bleeding vs stable, which one would you rather be if you were in the hospital. =D

 

2009/3/27 dujinfang <dujinfang@gmail.com (dujinfang@gmail.com)>

[anthony.minessale at g...]

We've made no attempts to add any optimization flags on unix to date.
We use the defaults and always build debug binaries.

When we get some spare time we might go back and turn them on but so far
we don't have much of a need to.


On Fri, Mar 27, 2009 at 1:56 PM, Gabriel Kuri <gkuri@ieee.org (gkuri@ieee.org)> wrote:

[gkuri at ieee.org]

ok, thanks, that pretty much amounts to the gcc defaults. perhaps I
should recompile FS with those defaults and see if the current jira I
have open goes away ...

Gabe

Brian West wrote:

[gkuri at ieee.org]

Thanks, Anthony. I guess that answers my question, the build process
doesn't use whatever is configured locally on the system.

Gabe


Anthony Minessale wrote:

[dujinfang at gmail.com]

Thanks.

I was thinking created a new server with Ubuntu intrepid, seems I'd like back to Hardy.
Even on hardy the default libtool is version 2. replace to libtool 1 should be easy as I mentioned before.



On Mar 28, 2009, at 3:04 AM, Anthony Minessale wrote:

[hads at nice.net.nz]

On Sat, 28 Mar 2009 15:22:23 dujinfang wrote: