[arch-dev-public] libusbx as replacement for libusb
Hi got this feature request: https://bugs.archlinux.org/task/29999 - fedora ditched the libusb usage in favour of the libusbx project. http://www.libusbx.org Shall we move to this too? greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org
Am 25.05.2012 14:56, schrieb Tobias Powalowski:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project.
Shall we move to this too?
Questions not answered in the bug report: 1) What is an actual issue solved by libusbx that is present in libusb? (by Allan) 2) Is it a drop-in replacement that is API-compatible or (better) ABI-compatible. ad 1) I guess there are such issues, otherwise the libusbx people wouldn't have been angry enough to make a fork. This point requires research. ad 2) If it is ABI-compatible, there is no harm in switching right now, as libusbx = libusb-1 + active development + more bugfixes. If it is only API-compatible or partially incompatible, we would need a strong reason to actually switch.
On Fri, May 25, 2012 at 3:18 PM, Thomas Bächler <thomas@archlinux.org> wrote:
Am 25.05.2012 14:56, schrieb Tobias Powalowski:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project.
Shall we move to this too?
Questions not answered in the bug report: 1) What is an actual issue solved by libusbx that is present in libusb? (by Allan) 2) Is it a drop-in replacement that is API-compatible or (better) ABI-compatible.
ad 1) I guess there are such issues, otherwise the libusbx people wouldn't have been angry enough to make a fork. This point requires research.
From what I gathered, the difference is currently minimal as the libusb maintainer backported most of the commits from libusbx. I assume this can not go on forever though...
ad 2) If it is ABI-compatible, there is no harm in switching right now, as libusbx = libusb-1 + active development + more bugfixes. If it is only API-compatible or partially incompatible, we would need a strong reason to actually switch.
It was advertised as a drop in replacement, I assumed this means ABI compatible, but this we must check. -t
Le 2012-05-25 09:25, Tom Gundersen a écrit : > On Fri, May 25, 2012 at 3:18 PM, Thomas Bächler <thomas@archlinux.org> wrote: >> Am 25.05.2012 14:56, schrieb Tobias Powalowski: >>> Hi got this feature request: >>> https://bugs.archlinux.org/task/29999 >>> >>> - fedora ditched the libusb usage in favour of the libusbx project. >>> >>> http://www.libusbx.org >>> >>> Shall we move to this too? >> Questions not answered in the bug report: >> 1) What is an actual issue solved by libusbx that is present in libusb? >> (by Allan) >> 2) Is it a drop-in replacement that is API-compatible or (better) >> ABI-compatible. >> >> ad 1) I guess there are such issues, otherwise the libusbx people >> wouldn't have been angry enough to make a fork. This point requires >> research. > From what I gathered, the difference is currently minimal as the > libusb maintainer backported most of the commits from libusbx. I > assume this can not go on forever though... > >> ad 2) If it is ABI-compatible, there is no harm in switching right now, >> as libusbx = libusb-1 + active development + more bugfixes. If it is >> only API-compatible or partially incompatible, we would need a strong >> reason to actually switch. > It was advertised as a drop in replacement, I assumed this means ABI > compatible, but this we must check. > > -t I was curious, so I tried the pkg from AUR [1]. My usb stuff still work, so I guess compatibility is not a issue. The choice will probably be done by considering which of the two projects is the most promising. Stéphane [1] https://aur.archlinux.org/packages.php?ID=59473
On Fri, May 25, 2012 at 2:56 PM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project.
Seems Debian have done/will do the same.
Shall we move to this too?
I am not very familiar with the background story, but from what I can gather from libusb-devel[0], it seems that the fork was done for a good reason, has lots of support, and no counterarguments have (as far as I could find at least) been presented by the libusb maintainer. To sum up, the main complaints were: libusb was not begin released for more than two years (it has now finally been released, after the fork happened), and the maintainer was seen as hostile to new contributions. Assuming the impression I got is correct, I would be in favor of switching. Cheers, Tom [0]: <http://libusb.6.n5.nabble.com/libusb-is-dead-long-live-libusbx-td5651413.html>
On Fri, May 25, 2012 at 2:56 PM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project. Seems Debian have done/will do the same.
Shall we move to this too? I am not very familiar with the background story, but from what I can gather from libusb-devel[0], it seems that the fork was done for a good reason, has lots of support, and no counterarguments have (as far as I could find at least) been presented by the libusb maintainer.
To sum up, the main complaints were: libusb was not begin released for more than two years (it has now finally been released, after the fork happened), and the maintainer was seen as hostile to new contributions.
Assuming the impression I got is correct, I would be in favor of switching.
Cheers,
Tom
[0]: <http://libusb.6.n5.nabble.com/libusb-is-dead-long-live-libusbx-td5651413.html> replaces=('libusb1' 'libusb')
Am 25.05.2012 15:21, schrieb Tom Gundersen: provides=('libusb') would that be ok in PKGBUILD? -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org
Am 25.05.2012 16:14, schrieb Tobias Powalowski:
On Fri, May 25, 2012 at 2:56 PM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project. Seems Debian have done/will do the same.
Shall we move to this too? I am not very familiar with the background story, but from what I can gather from libusb-devel[0], it seems that the fork was done for a good reason, has lots of support, and no counterarguments have (as far as I could find at least) been presented by the libusb maintainer.
To sum up, the main complaints were: libusb was not begin released for more than two years (it has now finally been released, after the fork happened), and the maintainer was seen as hostile to new contributions.
Assuming the impression I got is correct, I would be in favor of switching.
Cheers,
Tom
[0]: <http://libusb.6.n5.nabble.com/libusb-is-dead-long-live-libusbx-td5651413.html> replaces=('libusb1' 'libusb')
Am 25.05.2012 15:21, schrieb Tom Gundersen: provides=('libusb') would that be ok in PKGBUILD?
Would it be ok for you if i bring this to testing repository? -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org
On Sun, May 27, 2012 at 9:44 AM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Am 25.05.2012 16:14, schrieb Tobias Powalowski:
On Fri, May 25, 2012 at 2:56 PM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project. Seems Debian have done/will do the same.
Shall we move to this too? I am not very familiar with the background story, but from what I can gather from libusb-devel[0], it seems that the fork was done for a good reason, has lots of support, and no counterarguments have (as far as I could find at least) been presented by the libusb maintainer.
To sum up, the main complaints were: libusb was not begin released for more than two years (it has now finally been released, after the fork happened), and the maintainer was seen as hostile to new contributions.
Assuming the impression I got is correct, I would be in favor of switching.
Cheers,
Tom
[0]: <http://libusb.6.n5.nabble.com/libusb-is-dead-long-live-libusbx-td5651413.html> replaces=('libusb1' 'libusb')
Am 25.05.2012 15:21, schrieb Tom Gundersen: provides=('libusb') would that be ok in PKGBUILD?
Would it be ok for you if i bring this to testing repository?
+1 -t
On Sun, May 27, 2012 at 9:44 AM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Am 25.05.2012 16:14, schrieb Tobias Powalowski:
On Fri, May 25, 2012 at 2:56 PM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project. Seems Debian have done/will do the same.
Shall we move to this too? I am not very familiar with the background story, but from what I can gather from libusb-devel[0], it seems that the fork was done for a good reason, has lots of support, and no counterarguments have (as far as I could find at least) been presented by the libusb maintainer.
To sum up, the main complaints were: libusb was not begin released for more than two years (it has now finally been released, after the fork happened), and the maintainer was seen as hostile to new contributions.
Assuming the impression I got is correct, I would be in favor of switching.
Cheers,
Tom
[0]: <http://libusb.6.n5.nabble.com/libusb-is-dead-long-live-libusbx-td5651413.html> replaces=('libusb1' 'libusb')
Am 25.05.2012 15:21, schrieb Tom Gundersen: provides=('libusb') would that be ok in PKGBUILD?
Would it be ok for you if i bring this to testing repository? +1
-t
Am 27.05.2012 12:13, schrieb Tom Gundersen: libusbx is now in testing. greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org
On 01/06/12 17:19, Tobias Powalowski wrote:
On Sun, May 27, 2012 at 9:44 AM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Am 25.05.2012 16:14, schrieb Tobias Powalowski:
On Fri, May 25, 2012 at 2:56 PM, Tobias Powalowski <tobias.powalowski@googlemail.com> wrote:
Hi got this feature request: https://bugs.archlinux.org/task/29999
- fedora ditched the libusb usage in favour of the libusbx project. Seems Debian have done/will do the same.
Shall we move to this too? I am not very familiar with the background story, but from what I can gather from libusb-devel[0], it seems that the fork was done for a good reason, has lots of support, and no counterarguments have (as far as I could find at least) been presented by the libusb maintainer.
To sum up, the main complaints were: libusb was not begin released for more than two years (it has now finally been released, after the fork happened), and the maintainer was seen as hostile to new contributions.
Assuming the impression I got is correct, I would be in favor of switching.
Cheers,
Tom
[0]: <http://libusb.6.n5.nabble.com/libusb-is-dead-long-live-libusbx-td5651413.html> replaces=('libusb1' 'libusb')
Am 25.05.2012 15:21, schrieb Tom Gundersen: provides=('libusb') would that be ok in PKGBUILD?
Would it be ok for you if i bring this to testing repository? +1
-t
Am 27.05.2012 12:13, schrieb Tom Gundersen: libusbx is now in testing.
greetings tpowa
Should be provides=("libusb=$pkgver")
On 1 June 2012 10:20, Allan McRae <allan@archlinux.org> wrote:
Should be
provides=("libusb=$pkgver")
Also, conflicts=("libusb") is missing. error: failed to commit transaction (conflicting files) libusbx: /usr/include/libusb-1.0/libusb.h exists in filesystem libusbx: /usr/lib/libusb-1.0.a exists in filesystem libusbx: /usr/lib/libusb-1.0.so exists in filesystem libusbx: /usr/lib/libusb-1.0.so.0 exists in filesystem libusbx: /usr/lib/libusb-1.0.so.0.1.0 exists in filesystem libusbx: /usr/lib/pkgconfig/libusb-1.0.pc exists in filesystem -- Andrea
Am 01.06.2012 11:04, schrieb Andrea Scarpino:
On 1 June 2012 10:20, Allan McRae <allan@archlinux.org> wrote:
Should be
provides=("libusb=$pkgver") Also, conflicts=("libusb") is missing.
error: failed to commit transaction (conflicting files) libusbx: /usr/include/libusb-1.0/libusb.h exists in filesystem libusbx: /usr/lib/libusb-1.0.a exists in filesystem libusbx: /usr/lib/libusb-1.0.so exists in filesystem libusbx: /usr/lib/libusb-1.0.so.0 exists in filesystem libusbx: /usr/lib/libusb-1.0.so.0.1.0 exists in filesystem libusbx: /usr/lib/pkgconfig/libusb-1.0.pc exists in filesystem
fixed. -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org
participants (6)
-
Allan McRae
-
Andrea Scarpino
-
Stéphane Gaudreault
-
Thomas Bächler
-
Tobias Powalowski
-
Tom Gundersen