Hi Henrik,
Il 00.00 09/03/2005 Henrik Nordstrom ha scritto:
I don't see why such releases would be needed beyond the nightly snapshots which by definition is ALPHA. We certainly do not want anyone to package binaries of these as we know it won't work proper, and for people to be able to submit meaningful bug reports they need debug symbols and preferably source, both of which is not included in binary packaging. Also, we do not want to attrakt too many people who have trouble to use the debugger to extract stack traces on releases of this quality.
Yes, I agree, ALPHA is not a good idea.
We have only ended up in this situation (no new PRE release in 1.5 years) because Squid-3 has for a long period not received the attention it deserves making the number of known bugs accumulate above the theshold where PRE releases can be considered valid. The only way out of this is to focus on getting the known bugs fixes moving HEAD back to the quality level expected for a PRE release.
Probably the main problem is that we have a release that was named PRE, but it was (and now is still) not at the PRE source quality code level, causing some confusion.
I know that I am partly responsible for this situation to occur in the first place by not being able to forward-port my own bug fixes in 2.5 in a reasonable rate, causing Squid-3 to not only detoriate from it's own bugs but also from the large number of known and already fixed bugs still existing in Squid-3, and I am is very sorry for this but the squid-2.5 maintenance has simply drained all the time I have had available and a fair bit more.
But it is also true that if at least the current momentum can be kept then we are not far from reaching the quality point where a new PRE release can be motivated, but we need to at least get rid of the worst and most striking bugs first.
For me a 3.0 PRE release should have at least the same functionality of current 2.5, plus any 3.0 specific improvement. And as I know, in 3.0 we are still lacking a decent NTLM support, but some work is currently in progress on the SPNEGO side.
If we want reach a PRE quality point in a few time, we could target for PRE4 a NTLM negotiate support with no more challenge reuse and target for PRE5 full SPNEGO support.
I do think most users understands the difference between a PRE release and a release candidate, or at least to have some kind of understanding that there is a difference.
In my eyes
- nightly snapshots or CVS is the bleeding edge, or ALPHA releases if you prefer. These targets people with reasonable knowledge in how to build and maintain Squid and who is not afraid of getting their hands dirty while isolating bugs or to update Squid on a daily basis to get a problem fixed.
- PRE releases is milestones where the developers (us) is reasonably happy about the quality of the distribution, allowing people to start toying with the new features but knowing that it does not represent the final design and features may still be missing or behave differently, but having some level of assurance that it won't bite them too hard. For a such frozen release to make sense in terms of release maintenance it should be able to live for some time without needing huge amounts of patches to be useable.
- RC (Release Candidate) releases is short term releases in preparation for a STABLE release. Should in theory not be needed, but that's in theory..
I like the RC release concept: for peoples it's something different and better than PRE.
Regards
Guido
- ======================================================== Guido Serassio Acme Consulting S.r.l. - Microsoft Certified Partner Via Lucia Savarino, 1 10098 - Rivoli (TO) - ITALY Tel. : +39.011.9530135 Fax. : +39.011.9781115 Email: [EMAIL PROTECTED] WWW: http://www.acmeconsulting.it/
