On Mon, Sep 27, 2010 at 12:23 PM, Andreas Andreou <[email protected]> wrote: > +1 ... > > and not really an objection, but a chance to discuss some "related to > releases" > issues. > > - Am I right in saying that apart from the amount of time that has passed > since > the last stable version, there's no other pressing need to move to 5.3 (and > thus > pronounce 5.2 stable) ?
In many ways a GA release is a marketing issue, not a technical one. But perhaps part of that marketing issue should be an implicit promise to backport critical fixes (see below). > > - My reason for asking this is not to stall the process, but to better > understand (or > make us define) what is supposed to go in each version, and what just needs > to > wait for the next version. It feels that currently this just depends on timing > and i'm wondering if we're fine with such a process. If we are, then let's > just > enforce it better - and if we're not, let's discuss the alternatives. We keep talking about regular, small releases, but it doesn't quite happen. For 5.2, I purposely stepped back a bit to let 5.1 get some real exposure as a stable release, and I think that was a success on multiple levels. Firstly, 5.1 has been useful and stable. It has minor ommisions (many fixed in 5.2). Further, I saw that when I stepped back, you and others stepped forward, such as backporting fixes from 5.2 to 5.1. That is objectively quite healthy for the project. Yes, there were some snafus w.r.t. classpath asset security and so forth. Rather than spin wheels fixing many bugs in prior releases, I'd prefer to make each release fully, or nearly, backwards compatible to the prior release, so that it is easiest for end users to upgrade to the latest release. Is that a fantasy or reality? I think it varies from user to user, based on what aspects of Tapestry they've chosen to override and how. But, yes, realistically, we probably should be backporting more bug fixes. Perhaps our definition of a critical bug is one that must be fixed in the current and the immediately prior, release. The crux of the matter is who is willing to step forward and work on backporting bug fixes? With Tapestry operating as a true open source project (unlke many other projects, where a single development shop "owns" most of the key developers) its very hard to assign responsibilities to people and make them stick. I don't have a solution to that, and I'm not yet in a position to hire people and pay for Tapestry development (more than the amount of my salaried time, which I pay myself, to work on Tapestry). > > - Now, my other reason for asking this is that the current process > basically also > creates unmaintained (or lightly maintained) branches. Does it make > sense to officially > declare them as such? For instance, it's obvious that we didn't handle that > "classpath resources" problem elegantly and uniformally in all the > branches... i dont > think anyone wants to see that repeated. > Also and fyi, i've recently disabled 5.0 builds in hudson because they > were failing for > the last 6 months - the failure (just_HTML of TemplateParserImplTest) > is probably JVM > specific (i had looked at it back then) and we've since changed more than once > the template parser... so, it doesn't really look like anyone will > ever spend time > on that. > > - Finally, backporting features / improvements. Is anyone interested > or committed to > that? For which versions? I often see issues opened against 5.1 or 5.0 > but fixed only > for 5.2 - and that's just ok, it is the reality. Are we all fine with that? > > > On Mon, Sep 27, 2010 at 20:53, Ulrich Stärk <[email protected]> wrote: >> +1 >> >> On 27.09.2010 19:36, Howard Lewis Ship wrote: >>> >>> I'll be creating a 5.2.1 release shortly, then put it to the vote. >>> >>> Will anyone object if the vote is to "release 5.2.1 as a beta"? We >>> don't have a standard procedure for the alpha->beta transition. For >>> beta->ga, I will want to have the release out as a beta, then >>> retroactively vote it ga once it has proved itself in the wild. >>> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected] >> >> > > > > -- > Andreas Andreou - [email protected] - http://blog.andyhot.gr > Tapestry PMC / Tacos developer > Open Source / JEE Consulting > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > -- Howard M. Lewis Ship Creator of Apache Tapestry The source for Tapestry training, mentoring and support. Contact me to learn how I can get you up and productive in Tapestry fast! (971) 678-5210 http://howardlewisship.com --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
