Thanks for the info, this is helpful :)
On 10/4/07, Faidon Liambotis <[EMAIL PROTECTED]> wrote: > > Hello, > This is a update on the current status of Asterisk in Debian. > Apologies for the really long mail, it is targetted both to users and > maintainers :) > > I'm Ccing asterisk-users as a one-time thing; users that are interested > can subscribe to our list[1] for updates to prevent noise on a > non-Debian list. Please Cc pkg-voip-maintainer on replies. > > 1: http://lists.alioth.debian.org/mailman/listinfo/pkg-voip-maintainers > > sarge/etch status > ----------------- > About a month ago, I fixed all the long-standing knownn vulnerabilities > in both sarge/oldstable (1.0.7) and etch/stable (1.2.12). > The updates are present security.debian.org a Advisory has been released > (DSA-1358[1]), thanks to Debian's Security Team. > > These updates are fixing CVE-2007-1306, CVE-2007-1561, CVE-2007-2294, > CVE-2007-2297, CVE-2007-2488, CVE-2007-3762, CVE-2007-3763 and > CVE-2007-3764 (...). > > 1: http://www.debian.org/security/2007/dsa-1358 > > lenny status > ------------ > 1:1.4.11~dfsg-4 has been recently uploaded to unstable. > The previously mentioned block by the openh323 dependency which > currently fails to build in unstable (binutils bug: #440015) has been > workaround-ed (by having less strict shlibs in openh323) > > From our POV, it's a good candidate for lenny/testing. However: > - it depends on perl and net-snmp versions that are not present in > testing and are not in a shape to be there; we'll need new versions > from the respective teams. > - asterisk needs to go together with yate because of a shared libpri > dependency. However yate is being blocked[2] by gtk+2.0. > - more importantly, asterisk produces an Internal Compiler Error of GCC > 4.2 on hppa (#445336). Until it builds successfully there, it cannot > migrate to testing. > > 1: http://bjorn.haxx.se/debian/testing.pl?package=asterisk > 2: http://bjorn.haxx.se/debian/testing.pl?package=yate > > 1.4.12 > ------ > Digium released 1.4.12 the day before yesterday. I have committed all > the changes needed and we are now up to date. > Fortunately, many of our fixes that I reported upstream have been > merged. I have manually ported bristuff 0.4.0-test4 to 1.4.12; it needed > many changes compared to the previous upstream updates. > I will forward my changes to kapejod so that he can hopefully release a > new version. > > supplementary packages > ---------------------- > * asterisk-addons (-mp3, -mysql, -ooh323c) are finally present in Debian > and should be ready to migrate to lenny after Asterisk does. Digium > released a new version along with 1.4.12 and I will update this ASAP. > * asterisk-chan-capi, asterisk-spandsp-plugins, asterisk-oh323 had > recents uploads and all are in a good shape. > * I am going to drop rate-engine from the archive (#444712) since it has > no users, it wasn't released with etch, has open bugs for a really > long time and is unmaintained by upstream. > * I tried compiling chan_misdn together with the mISDN maintainer (Simon > Richter) and failed because of an mISDN API mismatch. > Need to take another look. > * asterisk-gui needs to be uploaded; Tzafrir? > * are we going to upload ARI? If not, we should drop it from our SVN. > * zaptel is in a good status and it's the only package from the suite > that is migrating to testing. Things TODO that come to mind are: a) > fixing a bug which results in /lib/modules/2.6.foo/modules.* files in > amd64 and b) evaluate a switch to OSLEC as the default echo > cancellator. Tzafrir is doing an excellent job on maintaining this > package by himself :) > * Right now, we are shipping asterisk-sounds-main which is the "main" > asterisk sounds in English in GSM format -- exactly as shipped in the > original tarball by Digium. Kilian, Tzafrir and me were pondering on > the idea of shipping separately all sounds as shipped by Digium in all > formats (besides WAV), each in a separate package. This should serve > our users better but has an obvious problem of size. This is not > decided yet. > > ABI issues > ---------- > Most -if not all- of these plugins build-depend on asterisk-dev i.e. use > Asterisk's development headers. These headers are tied to the ABI and > this can only be expressed in dependencies manually. > asterisk-chan-capi was compiled with 1.2 asterisk-dev, had a >= 1.2 > dependency but segfaults on 1.4 (#441237). There are currently no > similar problems that I know of. > However, we should expect more of these when we transition to 1.6 which > will most probably have a different ABI. > > I'm leaning towards a solution: > * Add a "Provides: asterisk-1.4" to asterisk. > * Replace "Depends: asterisk (>= 1.4.0)" (or similar) with "Depends: > asterisk-1.4" on all external modules. > This should help in *breaking*, dpkg-wise, the modules when a new > version is uploaded which in turn will prevent a new version from > entering testing until all plugins are recompiled. > > pushing our work upstream > ------------------------- > On the 1.4.11-1.4.12 cycle, I tried pushing all of our patches to > Digium's BTS (mantis). This has worked well since they're quite > responsive (contrary to our secondary upstream, Klaus-Peter Junghanns...). > I have began adding comments to all of our patches with either: > * the Digium ticket number > * a temporary "should be forwarded upstream" tag > * what's wrong with a patch that makes it unsuitable for forwarding it > I find this a good "policy" that we should probably adopt for all > pkg-voip's packages. > > getting upstream's work back :) > ------------------------------- > Digium is following an old-style Linux release model (odd-even > versions). Unfortunately, this leads to long delays between interesting > features or non-trivial fixes. > This, combined with our long release cycles can result in *huge* delays. > For example, we shipped etch in April with Asterisk 1.2 and Asterisk 1.4 > was feature-frozen for already 3 months at that point. > > We are currently shipping func_devstate (it is maintained for 1.4 > out-of-tree by its author, Russell Bryant) and I have just commited a > backport of the libcap patch so that asterisk can set the ToS of IP. > > I have also backported chan_mobile for asterisk-addons; I'm going to buy > a Bluetooth adapter and test it before uploading, however. > > Backporting stuff from trunk may be error-prone and is not easy to draw > a line of which stuff we should backport. > I'm open to suggestions on other modules that may have sense in > backporting. > > bugs > ---- > Our bug count is currently: > * Status > - 8 Outstanding > - 3 Forwarded > * Severity > - 2 Important bugs > - 2 Normal bugs > - 7 Wishlist items > This is probably the lowest bug count in ages, partly because testing > has still 1.2. > > I have begun forwarding upstream bugs to mantis. This has already been > proven fruitful: #353227 is open for 1 year and 230 days; I forwarded it > on 2007-09-21 and got a proposed patch back on the very next day! > > I'm expecting more bugs when asterisk migrates in testing; we can only > cope up with them if we tag them properly and forward the upstream ones > quickly. > > helping out > ----------- > You can help by testing our packages and reporting bugs if found. > Besides frequent uploads to unstable, there is buildserver that has > revision snapshots for etch, lenny, sid[1] and edgy, feisty and gutsy[2] > > Any feedback is welcome; suggestions on how to improve the packaging, > bug reports, backport requests, suggested out-of-tree patches for > inclusion etc. > > 1: http://pkg-voip.buildserver.net/debian/ > 2: http://pkg-voip.buildserver.net/ubuntu/ > (apt repositories) > > Best regards, > Faidon > > _______________________________________________ > --Bandwidth and Colocation Provided by http://www.api-digital.com-- > > asterisk-users mailing list > To UNSUBSCRIBE or update options visit: > http://lists.digium.com/mailman/listinfo/asterisk-users > -- With Best Regards, *************************** Sarfaraz Chougule http://www.sarfu.co.in ***************************
_______________________________________________ --Bandwidth and Colocation Provided by http://www.api-digital.com-- asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
