Hi all, Sorry for being quiet so far regarding this issue, but I've been too busy with other real-life subjects (besides, it's World Cup time :-).
Anyway, I've read all messages and will try to write a 'condensed' reply of all pertinent issues, plus a couple of statements summarizing them. As such, it's going to be a long email - so, if you don't have the patience to read all replies (I wouldn't :-), jump straight to the end... On Jun 6, 2006, at 6:13 PM, Rahul Akolkar wrote: Yup, there clearly is developer/community interest towards the formation of this project. I second Rahul's comment here. In fact, the proposal started as an effort from Hen to emancipate some projects (Cactus and JMeter) out of Jakarta, but there are many other projects interested to join the new TLP (I will talk more about those projects later). Plus, there is a chance to rejuvenate some existing projects by sheer proximity to newer projects with active developers (amongst other things). This is another good point: one motivation for the TLP is to bring momentum back to some dormant projects (like Cactus). I'm aware this motivation could be dangerous (we could, for instance, end up with a dormant TLP, which is worse than a dormant sub-project), but I'm still confident it's worth a try. Per the umbrella concern, the question then becomes what -- if any -- are the mitigating factors that can address such a concern with regards to this proposal. Ok, that makes sense: such mitigating factors should be on our proposal for the next meeting (I will bring them back after the replies). On Jun 6, 2006, at 9:47 PM, Henri Yandell wrote: It's more in our court to come up with something to convince them I think. Ok, let's try to come out with concrete arguments for the next meeting. Mostly I think we need to detail the cross-ASF interest in the idea. Otherwise Jakarta Test Ok again, let's do that. So far, I can list the following: - Struts developed some testing artifacts also used by MyFaces - WebWork - which has 'merged' into Struts - seems to have some testing stuff which could be migrated to the TLP - Cactus is (I believed) used by other JavaEE related projects (like Geromino and Struts) - we (Rahul and I) have been contacted in private by committers of other ASF projects (like Tomcat and Struts) willing to donate some code to the new TLP - as mentioned in previous messages, there are many other examples of testing artifacts spread across ASF projects that could be migrated into the new TLP. Of course, each case should be analyzed in particular, as not all of then might be suitable for the TLP, but the point is that we have a 'market' for the TLP. On Jun 9, 2006, at 2:50 AM, Phil Steitz wrote: I would also like to understand exactly what the problem is I think the problem is the fear that the TLP, as an umbrella project, grows up in an unorganized way and becomes more of a problem than a solution. and what mitigating steps may be possible. One such step is to have well defined rules on how an existing project would be accepted in the TLP. For instance, the proponent should 'prove' that the new project would aggregate value to the TLP, either technically and/or by bringing 'development momentum'. Another step (related with the previous one) is to define the incubation/sandbox mechanism for such new projects in a way a little bit more rigid than the regular incubation process. In particular, I would very much appreciate a definition of "umbrella" that allows Geronimo, Logging, Jakarta Commons, DB, XML, Web Services and Struts, but somehow disallows Testing. As others have already pointed out (sorry again for the delay on the reply :-), that definition is not a consensus. Anyway, I think Geronimo and Struts could be risked off the umbrella moniker, as they are focused in a concrete product and not on a generic concept (like the others). Besides, Jakarta Commons - which is the most 'problematic' umbrella, as it's very broad - is not an TLP, but a Jakarta sub-project (well, that's another issue...). On Jun 9, 2006, at 8:55 AM, Jesse Kuhnert wrote: It makes sense that people want to be careful about a tl subdomain. Some of the projects you mentioned are fairly staple diets to a good majority of development projects. (ie struts/logging/xml/commons/etc) . Yes, that's sort of what I meant previously (sorry for the redundancy). Maybe we (ASF) need a formal definition of what makes a TLP: for instance, it could be either a major project with some minor sub-projects (like Struts, Tomcat, Geronimo, Maven, Tapestry, etc..) or the (in) famous 'umbrella' (like Jakarta, DB, XML, Logging, Commons and hopefully Testing). In other words, we would be treating the umbrellas as 'first-class TLPs', and not some sort of 'failed experiments' (please note that I'm not affirming the aforementioned projects are failures, much the opposite. It's just a lack of a better expression :-( What would go into testing.apache.org? I'm all for it as testing in general has to be good thing As we mentioned before, anything (testing-related) could go into it, although the main focus would initially be Java-related projects. Note, though, that this 'bias' through Java is because of the TLP roots and - most important - the background of the initial PMC members - the TLP per se will be language-agnostic. and there is potential for all sorts of shared support if it is made easy to contribute into and collaborate on...Esp in the web based items world. I agree. There are many aspects of web development that are lacking better testing support, like JSF, AJAX, Portlets, etc.. (I'm not saying there isn't testing support for these areas, just that the existing support is not well consolidated yet). At the same time, if it's not substantial looking enough it could ~potentially~ be viewed as a negative thing. I think it is substantial looking enough; the negative thing could be the opposite, i.e., being too broad. On Jun 9, 2006, at 2:34 PM, Rahul Akolkar wrote: I think we have said that any codebase (beyond the seed set) that comes into Testing will have to have: a) A developer base willing to invest energies b) Existing community (for any that get incubated in -- ofcourse, that will be overseen by the Incubator as well) c) Binding support that will look for (a) and (b), amongst other things Do we agree on this? Any other comments? I agree, although we should improve the definition of 'developer base' (it could be a single developer, couldn't it?). Also, we need rules on what would happen with the developer base, i.e., if they should automatically become committers or not (although I think that's something we can define later in the inlaws of the project - I don't think it's relevant right now). Beyond that, IMO, it comes to having the board understand that we are as discerning of the umbrella concern as they are. And that the benefits seem to outweigh this concern. I agree, specially the part of the benefits outweighing the concerns. But we need bring them concrete arguments for that balance (I will list some in the end of this message). suspect that is only because we haven't talked much about Testing outside Jakarta (AFAIK). That's another good point, the discussion has been 'restricted' mostly to the Jakarta folks and the board. It made sense when the focus of the proposal was Hen's initiative to graduate some projects out of Jakarta; now that the issue is whether or not such testing umbrella is a Good Thing, it makes sense to spread it out of Jakarta. What about starting a thread on the community list? Towards the feedback in the initial email in this thread then, IMO, this doesn't feel artificial (perhaps we need some more clarification what that meant in the first place). Agreed. On Jun 9, 2006, at 2:48 PM, Rahul Akolkar wrote: If we're going to stand, we are going to do it on the basis of the merit of our proposal and the community support for it, rather than some sort of comparative analysis. Another nice point :-) I would also add that the community support endorses it's merit; we just make it clear for the board... On Jun 9, 2006, at 3:05 PM, Jesse Kuhnert wrote: You may want to pull in someone from the webwork/struts (I don't know what it's called right now) project. Specifically - Patrick Lightbody is pretty active in the area of testing so getting him to dump in thoughts might help. I'm not sure if it's a good idea right now. I mean, of course, such contribution will be welcome once the project is established, but I think for now we should focus on prove the merits of the proposal and we have already enough contenders for adoption. OTOH, bring them aboard now might help to show that the proposal has community support... You also have ibm and what they are doing in atf-dev @ eclipse. By speaking of Eclipse, Cactus is already going to be included on WTP in some way, so in the future (again, once the project is established), we could try a closer relationship with them. http://jrex.mozdev.org/, but I've not tried it.) I know a couple of the devs there have expressed definite interest in anything related to javascript/web/XHR unit testing solutions. (Javier Pedemonte/Adam Peller @ Ok, let's save these references for the future... On Jun 10, 2006, at 2:06 PM, Rahul Akolkar wrote: I only spend my whole day at the big blue, I had no such plans (and Now that you mentioned IBM, one thing that could help in the long term is getting IBM (or other company) to 'sponsor' the project. I mean, one pattern I have realized by watching/participating in many ASF projects is that those projects that have a company behind it has more chances to suceed - see, for instance, Geronimo, Maven, Derby and Tomcat. So, if IBM (or other companies) could allocate some people to the project (either their own employees or hiring some existing committers), that would help a lot (I know this might sound too ambitious for now, but as you spend the whole day at it, you might get us some help in the area :-) On Jun 9, 2006, at 6:33 PM, Henri Yandell wrote: time. My Internet time dives, and there are board elections happening in a couple of weeks so the board who pick up the tabled motion are not the same as the board who tabled it. I'm a little bit lost here regarding the dates: are the elections going to be held before the next meeting? Or should we focus our efforts on the next meeting (on the 21st) so the TLP creation is (hopefully) approved by the current board? You've got the right idea though, show the value of a testing.apache.org community to the ASF, and the wide interest in it. Yes, we got it. And thank you for all the fish, I mean, help, so far... On Jun 10, 2006, at 12:18 PM, Phil Steitz wrote: That's not what I meant. If the objection is "this looks like an umbrella, and umbrellas are evil" it is fair and reasonable for us to ask what exactly is meant by an umbrella so that we can address the specific concerns directly. Agreed. As I mentioned before, the board should come to a formal definition of the umbrella moniker, so we should include it on the next meeting's agenda. On Jun 10, 2006, at 2:06 PM, Rahul Akolkar wrote: I think we can go by the mantra "If we build it, they will come" I think so too, Wayne :-) OTOH, it will help to "socialize" Testing within the Apache community, so if you participate in projects that may be interested (or are attending ApacheConEU, for instance), please "spread the word". Maybe a quick BOF or something like that would be a good option (unfortunately I'm not going to ApacheConEU though, although I will try to go to the US conference). On Jun 11, 2006, at 7:10 PM, Phil Steitz wrote: Thanks for clarifying this, Hen. I thought PMC chairs always attended the meeting and that they were otherwise closed. Sorry for the BTW, that brings another question: does the chair has to be a member? If so, we must appoint another one, as I'm not. OK, I will plan to join the meeting and do my best to get a clear picture of what the board is looking for. That would be great, thanks. On Jun 12, 2006, at 7:36 PM, robert burrell donkin wrote: we don't have the answers. we may not even know the questions. The answer is 42 :-) we need to people to ask 'why?' (so please don't stop) Ok, in this case we should ask again Phil's question: what's the definition of an umbrella and why some projects (like DB, XML, WS, logging, etc..) can exist and Testing somehow can't? coming back to henri's comments: the ASF prefers self-organisation. reorganisations are much more likely to be approved if it's the committers involved who are pushing for them. if the communities are effected are strongly in favour then this has great weight. This is counting in our favor then, as the reorganization originated from the (Jakarta) committers and there is a large interest from the community (for instance, the proposed PMC has members that were part of Jakarta but not committers for Cactus or JMeter). So, now that the replies are over, let me try to consolidate the information, so we can have a plan for next steps: 1. We will try to pass the resolution (of the Testing TLP creation) on the next board meeting (on 21st?) and Phill Steitz will be our proxy. As he is not directly involved in the process (at least not yet :-), we must come out with whatever he should present in the meeting; 2. We understand the board's concerns about an umbrella, but we would like a formal definition of what an umbrella is, i.e., if it's allowed to have new sub-domains as umbrellas and what rules they should follow; 3.TAO (testing.apache.org) will initially contain only the JMeter and Cactus projects, migrated from Jakarta 4.TAO will be open to any testing-related project, independent on the technology or language involved. But it will have guidelines on whether a new project would be accept on it, and these guidelines will focus primarily in the ability of the project to maintain itself (i.e., the developer base, community involvement, etc..). These guidelines are also the main mitigating factor against the 'kitchen-sink umbrella concern'. 5.It's not clear yet how new projects will be accepted (although I guess it's the regular incubator process), but that's something we can discuss later, once the TLP is created and in place; 6.We are confident there is a growing interest in the community (specially among other ASF projects) for this TLP; a few people already manifested such interested. We are also aware this can be both a good and bad sign (as it could lead to the umbrella bloat), but we believe the benefits outweighs the concerns; 7.So far this discussion has been 'restrict' to [EMAIL PROTECTED] list and the board; we could 'spread the word' through the [EMAIL PROTECTED] list and ApacheConEU, but I'm not sure if it's necessary at this point (as mentioned on item 6, there is real interest outside of Jakarta already ) I think these statements are a good start for the next meeting's proposal - could someone write an wiki entry for it (or even update the current resolution)? I'm traveling until Sunday and my internet connection is pretty bad here, so it would be hard for me to do it... []s, -- Felipe --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
