Hi Juha,
Juha Heinanen wrote:
hi bogdan,
i'm trying to understand this in order figure out what to do in the future.
- the need for a new reliable release - delayed or low-quality releases
are affecting the credibility of the project
are all the open dialog and tm related issues on openser tracker fixed in opensips?
shortly yes, and not limited to. but that issues are just couple of the problems preventing the openser release - and are reflecting only the visible technical part. We should not overlook all the management for preparing the release and how the whole process is handled (or miss-handled).
- degradation of the project quality - testing, performance measurements
and code quality control are overlooked
i have been actively testing and measuring trunk version of openser since january. i assume that many others have been doing the same. could you be more specific?
Exactly this is the issue - people assumed that maybe somebody else is doing something about testing and measurements, but there is no consolidation on project level (only at personal level). In most of the cases, just assuming, we end up in actually having nobody doing this. In the last releases, it happend several times to discover some obviously problem right after the release.
- lack of the control, management and coordination of the project -
critical project issues could not be handled or solved
are you referring the local route issue? it was finally solved, but perhaps not to everybody's liking, which is normal in any democratic process.
not really - I do not refer to local route issue. From management point of view, simply review the board list to see what I'm referring at (how the critical issues related to the very existence of the project were handles - like changing the project name). Also "cawboys" like decisions - like revoking my project admin rights with no noticed just because I start working on OpenSIPS - and without expresing yet my position regarding OpenSER (if I still want to be involved or not). Technically speaking also, the decisions are taken in the "wild - cawboy" manner.
- a 6 months release cycle, but with no compromise for the technical
issues
what does that mean?
it means we will try to do a release at each 6 month, but no release will be done just because of the deadline while technical problems are still pending. Also a better pre- scheduling of the work will make more efficient the release progress. We have to admit there is a need for a better control and management.
- extensive testing and performance measurements before each major
release
how is that managed?
during the release cycle, the important steps like testing (join effort) are mandatory
- technical management with a strict developer hierarchy and developer
cooperation to prevent deadlocks or un-availabilities
how does this hierarchy currently look like? i didn't find anything about it on opensips web site. is there a technical "president" or "king" at the top of the hierarchy?
no :). the hierarchy is at follows (I will also upload it on the web): 1) core developers - they take design and arch decisions (the guys with the big picture), agree on accepting new modules; they have the last word in the technical issue (if not already solved below). 2) module maintainers - they maintain certain modules and take care all decision regarding their modules. For changes out of their scope, they need to address to level 1) 3) contributors - (see http://opensips.org/index.php?n=Development.Contributing)
any issues will be solved bottom up - if a current level is not able to solve it, it will be pushed to the upper level.
Regarding the "un-availability" - each module will end up having multiple (compatible) maintainers, so some code will not end up depending of the availability of only one person.
- intellectual honesty by having decisions taken on value-based
criteria and arguments.
how does this work with the strict 6 month release cycle?
as said, the 6 months release cycle is targeted if there are not technical stoppers - like fixing thinks or the need for a large work that may exceed 6 months. The critical need is to have pertinent decisions (in any context), based on value and arguments and not just baseless opinions
i don't claim that openser project has worked flawlessly, but the description in opensips announcement does not yet convince me either
I think no announcement can convince when comes to large projects. How things will progress (or how things are already progressing) will convince. Personally I put more trust in actions than in words.
Regards, Bogdan