[Freeswitch-dev] version numbers

Tom Parrott tomp at tomp.co.uk
Sat Nov 12 17:30:58 MSK 2011


I would be interested to know how you and the other Freeswitch 
developers go about deploying new 'versions' of the code.

Traditionally Linux uses various flavours of package management, such as 
RPM and APT that takes an upstream source tarball, combines it with 
patches and any other vendor supplies changes and then makes a package 
that can be installed on multiple servers.

At Infinity Tracking we have been using Freeswitch in production for 
over 6 months now and it has been extremely reliable.

We run several servers in a cluster, so it is important that we can 
manage code deployments effectively.

We use RPM and Puppet to ensure consistent package versions.

With every other piece of software in the stack we use; MySQL, PHP, 
Apache, Nginx, Memcached, Redis to name a few they are released as 
tarballs or at least versioned tags in git/svn.

This enables us to pull down either a tarball or extract a specific tag 
version and from there build a package with a version number.

With Freeswitch we have to 'git pull' the latest head, and then 
increment our build number, e.g. 1.0.7-12, however our version number of 
the package is internal to us, and so is of no use to other people for 
reporting bugs against.

I understand your concerns about the rate of development and how you 
don't want to get people stuck on older versions, and we don't expect a 
tagged version to instantly deployable without internal testing 
ourselves, just like git head wouldn't be.

I was wondering is there any possibility of having something like a 
'nightly' build version that is tagged in git that people could use as a 
common reference point?


Join us at ClueCon 2011 Aug 9-11, 2011
More information about the FreeSWITCH-dev mailing list