On Thu, Jul 5, 2012 at 1:18 AM, Wander Lairson Costa wrote: > 2012/6/30 Tormod Volden: >> On Thu, Jun 28, 2012 at 12:15 AM, Wander Lairson Costa wrote: >>> 2012/6/27 Tormod Volden: >>>> On Wed, Jun 27, 2012 at 12:31 PM, Wander Lairson Costa wrote: >>>>> Well, at worst, libusbhost would be another backend, based on the >>>>> libusb backend. Currently, due to my lack of free time, I am not >>>>> working on PyUSB (except by supporting mailing list, merging patches >>>>> and fixing simple bug fixes). If you provide the patch, I can merge it >>>>> in the current code (after a review, of course) , or if you are not >>>>> going to do that, you can create a new ticket for it, and at the right >>>>> time I make the implementation. >>>> >>>> Hi Wander, >>>> > > Hi Tormod, > > Sorry for long delay in responding the thread, but I have been in busy week... > >>>> Incidentally, if you are not working so much on pyusb for the time >>>> being, maybe you can tag a 1.0-alpha3 release soon since there are a >>>> number of fixes accumulated since 1.0-alpha2? >>>> >>> >>> I could not remember right now any critical fixes since alpha 2, but >>> if there is one that is heavily affecting you, I can tag the release >>> with no problem. >>> >> >> I have no critical fixes in mind, but I see that these alpha releases >> get picked up in various distributions so it is good to keep them >> coming. And I think 0.4 compatibility is very important for adoption, >> so compatibility fixes is always good to get out. >> > > 0.4 compatibility is indeed very important, but I think people won't > start using PyUSB 1.0 because of that, since they already have a true > and stable 0.4 API. libusb 1.0? They can get it through libusb-compat.
I think people will start using it when you tell them it is ready to be used :) Now the situation is that application programmers won't use 1.0 because it is not a stable API and it is not generally available in distributions. The fact that some distributions do have it must mean that they think it is already very useful. > >>>> Is there something in particular missing for a 1.0 release that the >>>> community can help out with? I checked Trac >>>> http://sourceforge.net/apps/trac/pyusb/report/1 but could not find any >>>> obvious blockers. >>>> >>> >>> Well, OpenUSB support is not ready (but since OpenUSB is not an >>> widespread library, it is not a showstopper for 1.0 release), >>> isochronous transfers (for libusb(x) 1.0 and libusb-win32) and I >>> didn't decide if libusbk will be in 1.0 or delayed to after stable >> >> Did this work in 0.4? As soon as there seems to be no regressions from >> 0.4 I think you should consider pushing 1.0, even if there are >> outstanding new features. Of course the API for existing functionality >> must be set in stone. >> > > IMO, there is no point in releasing a new version which changes Changes nothing? To quote the pysub web page :) Support for libusb 0.1, libusb 1.0 and OpenUSB. Easy API to communicate with devices. Support for custom library backends. Isochronous transfer type support. 100% written in Python by ctypes. It runs on any Python version >= 2.4 (this includes Python 3). > nothing. Also, I see no good point releasing 1.0 as stable with > possibility of radical changes in the API yet. If you plan radical changes in the API do not release it. Discussion over. But it sounds like you are not going to work so much on it in the near future, so maybe those changes can wait until 2.0? But giving the developers the hard choice between 0.4, 1.0 and 2.0 is not good either. It is already too much with 0.4 vs 1.0, libusb 0.1 vs 1.0 etc in my opinion. So if you delay 1.0 a bit longer to get the API you want, that sounds good to me. > >>> release. We also need better error checking (some errors are reported >>> in a too low level way in my opinion) and guarantee it works fine in >> >> Again, is this still better in 0.4? >> > > No. Even so, I haven't tested 0.4 API so well that I am comfortable to > say that compatibility layer stable. I agree this should be well tested before any 1.0 release, > >>> all Python version >= 2.4 (including Python 3). That's is what I >>> remember by now. The fact that these stuff it not in the tracker is a >>> completely lack of care of mine. >> >> Some of the testing could maybe be coordinated on a wiki page? If >> there is a defined set of test cases, volunteers could test on >> different versions and in different environments. >> > > Actually the source code has regression tests. But you need a PICDEM > like board to test it. I am using a modified version of the libusbk > Benchmark firmware [1] to run the tests [2]. The need for special hardware is a turn-off for community testing, but I understand this is hard to get around. > >>> >>>> Releases attract users, developers and hopefully contributors :) Of >>>> course it must be ready for consumption to avoid a heavy support >>>> burden. >>>> >>> >>> I agree. >>> >>> As a final word, if someone is willing to keep going with PyUSB >>> development I am more than happy to add him/her as a project admin. I >>> cannot help with coding right now, but I can help by directing what >>> needs to done and, may be, how I imagine things should be done. >> >> A listing of what needs to be done, in the tracker or wiki, would make >> it easier to contribute. But people usually contribute when they have >> their own motivations for it :) >> > > Well, I am moving wiki and issue tracker to github because SourceForge > is shutting down Trac, then I will organize the pending tasks for 1.0 > stable release. Hopefully interested people will help to get there. Thanks. Will have a look. Tormod > > [1] http://code.google.com/p/usb-travis/source/browse/#svn%2Ftrunk%2FBmFW > > [2] https://github.com/walac/bmfw > > -- > Best Regards, > Wander Lairson Costa ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ pyusb-users mailing list pyusb-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pyusb-users
