From: Pete B. <pb...@gm...> - 2010-10-30 18:02:09
|
OK, I'm going to try to keep my reply short as well, and reply to various points in no particular order. 1. Peter, please look up what the E in RERO stands for. It's called RERO rather than RO for a very good reason. We've has a Windows code that has been usable for the past 9 months. If 9 months qualifies for early, one's gotta wonder what qualifies for late... 2. If you're not familiar with an area, learn to trust that your contributors are not trying to screw the project up or delegate. Else, while it may seem admirable that you try to have an compleet overview of everything, you're going to delay the whole thing to unreasonable delays, and that will not serve anybody. Besides, you can always integrate something and then review it once you've had a chance to become more familiar with it. That's the (apparently much unused) beauty of VCS's like git - you can remodel everything to your heart's contents. The only thing that warrants a great deal of effort about getting it right the first time is the API, which is not an issue with the Windows backend. For all the rest, not so much. 3. As Ludovic pointed out, if reviewers are silent, shouldn't that mean tacit approval? How long should the timeframe for review be? You might want to take some hints from the OpenOCD maintainers, and call "going once, going twice" over the run of a few days or weeks, to bring the review process to a reasonable length of time. 4. There remains some major uncertainties that have yet to be resolved with regards to the fate of the DLL generation and the inclusion of the MSVC project files. 5. Yes I do believe that frequent releases drive developers and contributors. You might want to read [1]. 6. On "Do you mean to blame Daniel and me for others not reviewing patches", I would blame you for not trusting that we're all working for the benefit of libusb, and somehow giving the idea that the Windows backend proposal is so poor that it needs extensive review. I firmly believe that going with the current DLL generation, the integration of a sample like xusb or providing the MSVC projects will greatly benefit your users, but your comment leave the impression that doing any of the above will be hugely detrimental for everybody, which, as others have pointed out, borders on the absurd. Regards, /Pete [1] http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/ar01s04.html |