Re: Hurd and the archive
On Mon, 2013-05-06 at 21:47 +0100, Steven Chamberlain wrote: Hi Samuel, On 06/05/13 21:35, Samuel Thibault wrote: Steven Chamberlain, le Mon 06 May 2013 21:20:57 +0100, a écrit : In that case would there be 150-200 RC-severity bugs introduced right away by its inclusion? I would rather say simply dropping them, as already requested in Bug#704477. And as I said a fair amount of these are actually already submitted as general FTBFS bugs or upgrade libtool bugs. If it's possible, yes outdated versions could be removed... and then look again at those figures. But it would need to happen pretty soon. Add to these numbers the large amount of bug reports _with_ patches not being handled during the long freeze time for Wheezy, see (note this list is not exhaustive, e.g. some patched packages are in experimental) bugs.debian.org/cgi-bin/pkgreport.cgi?tag=hurd;users=debian-hurd@lists.debian.org 1 serious 49 important 4 normal (1 wontfix) 2 wishlist 10 forwarded important 3 pending upload important etc. Let's do the calculation of the coverage percentage when these bugs have been attended to (and the outdated ones removed as above). Why do you expect anything to be different now compared when the freeze happened, _several months ago_, in zero time? -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1368109546.4595.14.camel@PackardBell-PC
Re: Hurd and the archive
Svante Signell, le Thu 09 May 2013 16:25:46 +0200, a écrit : Why do you expect anything to be different now compared when the freeze happened, _several months ago_, in zero time? Please avoid this level of language, it can't bring anything good to the discussion. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130509143916.gg8...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
On Thu, 2013-05-09 at 16:39 +0200, Samuel Thibault wrote: Svante Signell, le Thu 09 May 2013 16:25:46 +0200, a écrit : Why do you expect anything to be different now compared when the freeze happened, _several months ago_, in zero time? Please avoid this level of language, it can't bring anything good to the discussion. Maybe the wording was a little strong, sorry for that. The meaning of the mail was to give a link to facts, and give room for things to happen before making a decision. What's the reason for such a hurry? Wheezy has just been released, thanks everybody :) Maybe the maintainers should be given some slack as pointed out by earlier posts (not including the GNU/Hurd ones) -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1368118446.4595.33.camel@PackardBell-PC
Re: Hurd and the archive
Paul Wise, le Wed 08 May 2013 12:06:01 +0800, a écrit : On Wed, May 8, 2013 at 8:21 AM, Samuel Thibault wrote: We haven't worked on microcode upload, since we don't necessarily target non-freeness... So, someday the Hurd will need support for microcode upload. Well, I haven't said we shall not implement it :) Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508074539.gu6...@type.wlan.youpi.perso.aquilenet.fr
Re: Hurd and the archive
On Wed, May 08, 2013 at 01:07:42AM +0200, Samuel Thibault wrote: Neil McGovern, le Tue 07 May 2013 11:14:01 +0100, a écrit : On Mon, May 06, 2013 at 10:27:54PM +0200, Samuel Thibault wrote: We have not worked too much on the hardware support in the past months, so it is basically network board drivers from linux 2.6.32, and IDE disk support. I for instance installed it on my Dell D430, and network just works fine. Working on a SATA driver should not be a problem. I just haven't put it high on my TODO list, and have rather worked on the Wheezy release whenever I had time to. Basically: in about 1 months time, will I be able to install it with a default installation process, and have it working on: a) A HP DL360 or similar b) A Dell inspiron 660s or similar c) A Lenovo Thinkpad X220 or similar I don't know what ethernet driver these would need. 2.6.32 linux kernels already have e1000, 8139*, tg3 etc. drivers. This is the usual issue of not-so-mature systems, just like Linux had in its early days. How about things like wireless drivers, raid controllers, suspend/resume, power management etc? About disk support, I happen to have right now a few days of holiday with no RL plans (at last!), so I'll work on the SATA driver. Having it working within a month should just happen. But not tested - how about USB - did that ever get sorted? d) VMWare/VBox etc. This already works. Just tried it with vbox - as soon as I selected 'text install', I got a critical error and the vm stopped. For something to be accepted in testing, it should be in a releaseable state. This isn't something I can see happening for Hurd. Neil -- signature.asc Description: Digital signature
Re: Hurd and the archive
Neil McGovern, le Wed 08 May 2013 11:35:52 +0100, a écrit : About disk support, I happen to have right now a few days of holiday with no RL plans (at last!), so I'll work on the SATA driver. Having it working within a month should just happen. But not tested - how about USB - did that ever get sorted? We have not worked on it. How about things like wireless drivers, raid controllers, suspend/resume, power management etc? There are some wireless drivers for pcmcia cards (e.g. orinoco, hermes). No raid support. No suspend/resume or power management. I'm wondering: if I had spent time on these instead of working on Wheezy, I guess people wouldn't have been happy either. I wonder what I should have done at all. And when these get implemented, I guess we'll be asked for 3D acceleration, backlight tuning, memory hotplug, etc. etc.? d) VMWare/VBox etc. This already works. Just tried it with vbox - as soon as I selected 'text install', I got a critical error and the vm stopped. I don't have this issue at all, things just go fine here with both the other/other template and the Linux/Linux template. This message comes from vbox I guess (there is no such message in Mach or the Hurd), so I'd tend to think virtualbox has some issues in your setup. For something to be accepted in testing, it should be in a releaseable state. Which we haven't seen very precisely defined still. Or at least we have this criterium: “Are machines available to buy for the general public?” Which I believe is fullfiled. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508143303.ga8...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
On Wed, May 08, 2013 at 04:33:03PM +0200, Samuel Thibault wrote: Neil McGovern, le Wed 08 May 2013 11:35:52 +0100, a écrit : But not tested - how about USB - did that ever get sorted? We have not worked on it. How about things like wireless drivers, raid controllers, suspend/resume, power management etc? There are some wireless drivers for pcmcia cards (e.g. orinoco, hermes). No raid support. No suspend/resume or power management. I'm wondering: if I had spent time on these instead of working on Wheezy, I guess people wouldn't have been happy either. I wonder what I should have done at all. And when these get implemented, I guess we'll be asked for 3D acceleration, backlight tuning, memory hotplug, etc. etc.? No, just something that works for the majority of our users. I'm fairly sure things like SATA and USB is considered essential. d) VMWare/VBox etc. This already works. Just tried it with vbox - as soon as I selected 'text install', I got a critical error and the vm stopped. I don't have this issue at all, things just go fine here with both the other/other template and the Linux/Linux template. This message comes from vbox I guess (there is no such message in Mach or the Hurd), so I'd tend to think virtualbox has some issues in your setup. I installed virtualbox on a standard Wheezy system, with other/other and 2G ram. For something to be accepted in testing, it should be in a releaseable state. Which we haven't seen very precisely defined still. Or at least we have this criterium: I'll see if I can be clear: I will not be putting Hurd in testing in the next few months. Neil -- signature.asc Description: Digital signature
Re: Hurd and the archive
Neil McGovern, le Wed 08 May 2013 16:07:26 +0100, a écrit : No, just something that works for the majority of our users. I'm fairly sure things like SATA and USB is considered essential. SATA, yes. USB, less. d) VMWare/VBox etc. This already works. Just tried it with vbox - as soon as I selected 'text install', I got a critical error and the vm stopped. I don't have this issue at all, things just go fine here with both the other/other template and the Linux/Linux template. This message comes from vbox I guess (there is no such message in Mach or the Hurd), so I'd tend to think virtualbox has some issues in your setup. I installed virtualbox on a standard Wheezy system, with other/other and 2G ram. Same here (3.2.0-4-amd64 kernel, kvm supported by my CPU). No issue. For something to be accepted in testing, it should be in a releaseable state. Which we haven't seen very precisely defined still. Or at least we have this criterium: I'll see if I can be clear: I will not be putting Hurd in testing in the next few months. I'm afraid I'll still not be able to explain people when they ask me why hurd-i386 is not in testing, except vague reasons. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508151644.gc8...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
Sorry to clutter the mailing list. But as a long time lurker at the mailing list I would just like to thank the Hurd team for all your efforts! @Samuel, I totally agree with you on your points. I have been happily running Hurd on my (reasonably new) core 2 computer for a year. It wasn't harder to install than a standard linux distribution. If you can accept installing via a curses based interface. As I have understood it the Linux kernel starts to become much like Microsoft Windows. Just patch after patch on basically the same base. With better distribution this will attract more devs, get better hardware support and grow into something great! With a new foundation, not being locked to old practices (yet). It just seems counter productive to rely on a small team of devs to do the same work as the huge team that Debian/Linux has at its disposal at the moment before supporting them. The Hurd team will never keep up and be able to add more features. Whenever you install an open source distribution you better check your hardware specs. And no matter if you run linux or Hurd, there is always something not working as expected! That's just how things are and have been for the last fifteen years (since I started using linux). Just my two cents, (which I know count for nothing). Anyway, keep up the good work hurd-guys! /Peter Samuel Thibault sthiba...@debian.org wrote: Neil McGovern, le Wed 08 May 2013 11:35:52 +0100, a écrit : About disk support, I happen to have right now a few days of holiday with no RL plans (at last!), so I'll work on the SATA driver. Having it working within a month should just happen. But not tested - how about USB - did that ever get sorted? We have not worked on it. How about things like wireless drivers, raid controllers, suspend/resume, power management etc? There are some wireless drivers for pcmcia cards (e.g. orinoco, hermes). No raid support. No suspend/resume or power management. I'm wondering: if I had spent time on these instead of working on Wheezy, I guess people wouldn't have been happy either. I wonder what I should have done at all. And when these get implemented, I guess we'll be asked for 3D acceleration, backlight tuning, memory hotplug, etc. etc.? d) VMWare/VBox etc. This already works. Just tried it with vbox - as soon as I selected 'text install', I got a critical error and the vm stopped. I don't have this issue at all, things just go fine here with both the other/other template and the Linux/Linux template. This message comes from vbox I guess (there is no such message in Mach or the Hurd), so I'd tend to think virtualbox has some issues in your setup. For something to be accepted in testing, it should be in a releaseable state. Which we haven't seen very precisely defined still. Or at least we have this criterium: “Are machines available to buy for the general public?” Which I believe is fullfiled. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508143303.ga8...@type.youpi.perso.aquilenet.fr -- Sent from my Android phone with K-9 Mail. Please excuse my brevity.
Re: Hurd and the archive
Samuel Thibault sthiba...@debian.org (08/05/2013): I'm afraid I'll still not be able to explain people when they ask me why hurd-i386 is not in testing, except vague reasons. For the record, and speaking only for myself, I'd love to see a proper architecture qualification process, rather than Neil's “approach” to saying “no” to Hurd based on a few hardware considerations. Now, I think everyone having actually worked on the release would enjoy (and probably deserve) a few days/weeks of rest before resuming release duties, which include: - handling transitions (and fun has already begun), - preparing r1 (and many packages were punted with r0 approaching), - thinking about longer term things, like arch. qual., release goals, etc. Hopefully we'll manage to handle arch. qual. earlier than during wheezy's release cycle. But please allow the release team to lean back a few moments. Mraw, KiBi. signature.asc Description: Digital signature
Re: Hurd and the archive
Cyril Brulebois, le Wed 08 May 2013 17:52:13 +0200, a écrit : Now, I think everyone having actually worked on the release would enjoy (and probably deserve) a few days/weeks of rest before resuming release duties, which include: I was also thinking the same when I saw the issue popping up already :) We are already getting things to handle in the whole distribution... Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508160934.gh8...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
On 08/05/13 16:52, Cyril Brulebois wrote: Samuel Thibault sthiba...@debian.org (08/05/2013): I'm afraid I'll still not be able to explain people when they ask me why hurd-i386 is not in testing, except vague reasons. For the record, and speaking only for myself, I'd love to see a proper architecture qualification process, rather than Neil's “approach” to saying “no” to Hurd based on a few hardware considerations. The hardware suggestions were a little bizarre. Some servers with NICs requiring non-free firmware to work; a laptop that didn't have working NIC or Xorg on GNU/Linux until Wheezy; and a desktop whose hardware chipsets are unspecified. I guess Neil had in mind someone setting up a buildd, a developer wanting to debug an issue with a package on that arch, and maybe a potential user - using whatever they had available. But typically it is optimistic to think even GNU/Linux would work properly on some given piece of hardware. Things mentioned, like hardware RAID support, suspend/resume and power management are precisely the sort of things that can go wrong. Rather, if you have a preference of operating system it is often necessary to research what is supported and what is not, and then shop around. Therefore we could use a list of example systems already working with GNU/Hurd, listing what hardware/drivers they use, and some comments on how well it all works. Hopefully we'll manage to handle arch. qual. earlier than during wheezy's release cycle. But please allow the release team to lean back a few moments. Even if the decision could be deferred until after r1, Samuel could use some hints on best to spend time between then and now to try to address concerns. My idea would be to look at the (I estimate) 90+ packages that are out-of-date in sid on GNU/Hurd but not other arches, and were not already removed by #704477. This includes some important ones like gcc-4.8, tar, gdb. git, grub2; keeping up with jessie development would be quite difficult without these being up-to-date. Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/518a8d21.5050...@pyro.eu.org
Re: Hurd and the archive
Steven Chamberlain, le Wed 08 May 2013 18:36:33 +0100, a écrit : My idea would be to look at the (I estimate) 90+ packages that are out-of-date in sid on GNU/Hurd but not other arches, and were not already removed by #704477. This includes some important ones like gcc-4.8, gdb. gcc-4.8 was actually broken by a patch addition to fix gdb which was actually already backported to upstream 4.8 :) git is a very new one, previous uploads just built fine with the whole testsuite. tar trips on one very specific testcase about symlinks. The grub2 looks like a transient misbuild. I guess the (built) experimental version will get uploaded soon anyway. But yes, now that wheezy is out we can get these fixed. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508175304.gr8...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
On Mon, May 06, 2013 at 10:27:54PM +0200, Samuel Thibault wrote: We have not worked too much on the hardware support in the past months, so it is basically network board drivers from linux 2.6.32, and IDE disk support. I for instance installed it on my Dell D430, and network just works fine. Working on a SATA driver should not be a problem. I just haven't put it high on my TODO list, and have rather worked on the Wheezy release whenever I had time to. Basically: in about 1 months time, will I be able to install it with a default installation process, and have it working on: a) A HP DL360 or similar b) A Dell inspiron 660s or similar c) A Lenovo Thinkpad X220 or similar d) VMWare/VBox etc. Neil -- signature.asc Description: Digital signature
Re: Hurd and the archive
Neil McGovern, le Tue 07 May 2013 11:14:01 +0100, a écrit : On Mon, May 06, 2013 at 10:27:54PM +0200, Samuel Thibault wrote: We have not worked too much on the hardware support in the past months, so it is basically network board drivers from linux 2.6.32, and IDE disk support. I for instance installed it on my Dell D430, and network just works fine. Working on a SATA driver should not be a problem. I just haven't put it high on my TODO list, and have rather worked on the Wheezy release whenever I had time to. Basically: in about 1 months time, will I be able to install it with a default installation process, and have it working on: a) A HP DL360 or similar b) A Dell inspiron 660s or similar c) A Lenovo Thinkpad X220 or similar I don't know what ethernet driver these would need. 2.6.32 linux kernels already have e1000, 8139*, tg3 etc. drivers. This is the usual issue of not-so-mature systems, just like Linux had in its early days. About disk support, I happen to have right now a few days of holiday with no RL plans (at last!), so I'll work on the SATA driver. Having it working within a month should just happen. d) VMWare/VBox etc. This already works. Also, Xen support just works, which is not really like raw hardware support of course, but still better than full virtualization. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130507230742.gf6...@type.wlan.youpi.perso.aquilenet.fr
Re: Hurd and the archive
On 08/05/13 00:07, Samuel Thibault wrote: I don't know what ethernet driver these would need. Should probably start a GNU/Hurd hardware status page on the Wiki... Neil McGovern, le Tue 07 May 2013 11:14:01 +0100, a écrit : a) A HP DL360 or similar The chipset has changed a bit between generations: G1: eepro (Linux) or fxp (*BSD) driver, requiring non-free microcode G2: HP NC7780 aka BCM5701, tg3 driver (Linux) or bge (*BSD) G3: HP NC7781, same? G4: HP NC7782, same? G5: bce/bnx (*BSD) or bnx2 (Linux) driver, needs non-free blob? 004:00:0: Broadcom BCM5708 NetXtreme II 1000baseT Ethernet (ethernet network, revision 0x12) G6: HP NC382i aka BCM5709C, same driver b) A Dell inspiron 660s or similar Chipset isn't mentioned in their tech specs, perhaps varies. c) A Lenovo Thinkpad X220 or similar I've seen e1000e mentioned, supported in wheezy by Linux 3.2.x, but not by the version of the Intel drivers in squeeze: 00:19.0 Ethernet controller: Intel Corporation 82579LM Gigabit Network Connection (rev 04) d) VMWare/VBox etc. This already works. Also, Xen support just works [...] With network connectivity? If so which NIC do they emulate? (rtl8139?) What kind of storage bus is used? Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/51899557.5090...@pyro.eu.org
Re: Hurd and the archive
Steven Chamberlain, le Wed 08 May 2013 00:59:19 +0100, a écrit : d) VMWare/VBox etc. This already works. Also, Xen support just works [...] With network connectivity? Sure. If so which NIC do they emulate? (rtl8139?) They don't emulate a nic, it's ParaVirtualization. What kind of storage bus is used? ParaVirtualized, too. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508001618.gh6...@type.wlan.youpi.perso.aquilenet.fr
Re: Hurd and the archive
Steven Chamberlain, le Wed 08 May 2013 00:59:19 +0100, a écrit : Neil McGovern, le Tue 07 May 2013 11:14:01 +0100, a écrit : a) A HP DL360 or similar The chipset has changed a bit between generations: G1: eepro (Linux) or fxp (*BSD) driver, requiring non-free microcode We haven't worked on microcode upload, since we don't necessarily target non-freeness... G2: HP NC7780 aka BCM5701, tg3 driver (Linux) or bge (*BSD) tg3 should work then. G3: HP NC7781, same? G4: HP NC7782, same? G5: bce/bnx (*BSD) or bnx2 (Linux) driver, bnx2 is supported. needs non-free blob? ditto. c) A Lenovo Thinkpad X220 or similar I've seen e1000e mentioned, supported in wheezy by Linux 3.2.x, but not by the version of the Intel drivers in squeeze: 00:19.0 Ethernet controller: Intel Corporation 82579LM Gigabit Network Connection (rev 04) Ok. Could be a matter of upgrading the driver. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130508002120.gi6...@type.wlan.youpi.perso.aquilenet.fr
Re: Hurd and the archive
On Wed, May 8, 2013 at 8:21 AM, Samuel Thibault wrote: We haven't worked on microcode upload, since we don't necessarily target non-freeness... I'm not sure where Hurd is at with wireless drivers, but there is some GPLed WiFi firmware out there: http://linuxwireless.org/en/users/Drivers/carl9170.fw http://github.com/qca/open-ath9k-htc-firmware In addition, firmware-linux-free contains some freely licensed firmware, including some for Ethernet cards. I believe the freely licensed nouveau drivers for nVidia GPU cards have the ability to generate the needed GPU microcode. So, someday the Hurd will need support for microcode upload. -- bye, pabs http://wiki.debian.org/PaulWise -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/caktje6hz-k1n1ofoxwyl4cntc9rsdj1ibt4p_5q1chbzrfm...@mail.gmail.com
Re: Hurd and the archive
On Sun, May 05, 2013 at 05:07:13PM +0200, Joerg Jaspert wrote: So, release people: How likely is it that Hurd gets added to jessie? Within the next one or two months I mean, not maybe in a years time. :) I don't see it happening, to be honest. Neil -- signature.asc Description: Digital signature
Re: Hurd and the archive
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 OK, I have been significantly out of the loop for a while now but what do you base that on? What requirements are we still falling short on? Thanks, Barry deFreese On 5/6/2013 7:08 AM, Neil McGovern wrote: On Sun, May 05, 2013 at 05:07:13PM +0200, Joerg Jaspert wrote: So, release people: How likely is it that Hurd gets added to jessie? Within the next one or two months I mean, not maybe in a years time. :) I don't see it happening, to be honest. Neil -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.10 (MingW32) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlGHsfYACgkQ5ItltUs5T36WwgCguHojqq0bKSU3CPoAbsrviQR7 CgMAn0Ij4Tdkei+BeLn+vzhMARplG474 =fS3d -END PGP SIGNATURE- -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/5187b1f7.9060...@verizon.net
Re: Hurd and the archive
On Mon, May 06, 2013 at 05:15:44PM +0100, Neil McGovern wrote: Percentage built, percentage up to date, and (as far as I know) a working port and installer for a modern desktop machine? Um, having read back the above, it may have sounded a bit more curt than I was expecting, apologies! Those are meant to be genuine questions. Blame the release etc... :) Neil -- signature.asc Description: Digital signature
Re: Hurd and the archive
On 05/05/13 16:07, Joerg Jaspert wrote: ...released together with all the others (probably as a technology preview)... So, release people: How likely is it that Hurd gets added to jessie? If added as a 'technology preview', what does that mean exactly? Would Hurd-specific RC-severity bugs stall a package's transition to testing? And would it be necessary to fix all Hurd-specific RC bugs to be able to release? From the view of maintainers I think that would be the deciding factor, because it could imply extra work. Not everyone sees the benefits of porting efforts (whereas I see it as excellent QA and promotes better software design, hence I'm in favour of inclusion). Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/5187de57.2000...@pyro.eu.org
Re: Hurd and the archive
On Mon, May 06, 2013 at 09:36:55AM -0400, Barry deFreese wrote: -BEGIN PGP SIGNED MESSAGE- OK, I have been significantly out of the loop for a while now but what do you base that on? What requirements are we still falling short on? Percentage built, percentage up to date, and (as far as I know) a working port and installer for a modern desktop machine? Neil -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130506161544.gf7...@halon.org.uk
Re: Hurd and the archive
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Fair enough and no offense taken. As I said, I haven't been contributing for quite a while now so they were more questions for myself than anything. Niels actually showed me the graph and it is a bit sadder than I had hoped. :( Though we do have an installer now and they have done some great work over the year. Thanks, Barry On 5/6/2013 12:17 PM, Neil McGovern wrote: On Mon, May 06, 2013 at 05:15:44PM +0100, Neil McGovern wrote: Percentage built, percentage up to date, and (as far as I know) a working port and installer for a modern desktop machine? Um, having read back the above, it may have sounded a bit more curt than I was expecting, apologies! Those are meant to be genuine questions. Blame the release etc... :) Neil -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.10 (MingW32) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlGH6RQACgkQ5ItltUs5T34VoACbBOFdUNXs3p6buSak9bRQ+0dO V7cAoN3eGJOoGa4oocx34jbypQeb0uoe =beuB -END PGP SIGNATURE- -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/5187e914.6080...@verizon.net
Re: Hurd and the archive
with all the others (probably as a technology preview)... So, release people: How likely is it that Hurd gets added to jessie? If added as a 'technology preview', what does that mean exactly? Note that the tech preview was a softening of a requirement to get added to wheezy. Which didnt happen... Would Hurd-specific RC-severity bugs stall a package's transition to testing? And would it be necessary to fix all Hurd-specific RC bugs to be able to release? Unless it is added as a (whats the term in the code? fucked_arch?) well, second class arch thats usually ignored, then yes, adding it will come with all the usual requirements of a release arch. Adding it and then keeping it out of the usual migration rules is asking for failure from the beginning, accumulating cruft. Not a way to go, IMO. From the view of maintainers I think that would be the deciding factor, because it could imply extra work. Not everyone sees the benefits of porting efforts (whereas I see it as excellent QA and promotes better software design, hence I'm in favour of inclusion). I would be in favour of including it if it actually would look like it could be as up to the task as all the rest of the architectures are. But it doesn't appear to me that it is that. -- bye, Joerg rvb Dafür hat Ubuntu nen kleinen. -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87ppx37vt9@gkar.ganneff.de
Re: Hurd and the archive
On 06/05/13 20:27, Joerg Jaspert wrote: Adding it and then keeping it out of the usual migration rules is asking for failure from the beginning, accumulating cruft. Not a way to go, IMO. In that case would there be 150-200 RC-severity bugs introduced right away by its inclusion? (Packages which FTBFS on hurd-i386 that are already 'out of date' in sid, counting Failed + Build-Attempted) : https://buildd.debian.org/status/architecture.php?a=hurd-i386suite=sidnotes=out-of-date This was the best figure I could think of to quantify the 'burden' of a particular arch being included in testing. For comparison, kfreebsd arches tally ~50, armel/mipsel ~50, ia64 ~60, amd64/i386 only 10-20. There is a lot of overlap though, e.g. fixing a kfreebsd build failure may fix hurd-i386. I found some mention/suggestion that for arch-specific issues, a 'technology preview' may be released even if some RC-severity bugs remain (though probably not when packages FTBFS); and relaxed criteria might be used during freeze and for stable updates: https://lists.debian.org/debian-bsd/2011/06/msg00365.html Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/518810a9.9050...@pyro.eu.org
Re: Hurd and the archive
Neil McGovern, le Mon 06 May 2013 17:15:45 +0100, a écrit : On Mon, May 06, 2013 at 09:36:55AM -0400, Barry deFreese wrote: -BEGIN PGP SIGNED MESSAGE- OK, I have been significantly out of the loop for a while now but what do you base that on? What requirements are we still falling short on? Percentage built, This has of course not changed since the Wheezy freeze 10 months ago: between 77% and 78%. Which have been rebuilt BTW. In the meanwhile we have pending fixes in various packages and glibc which should raise that, but we didn't push them during the freeze. It also has to be considered that this is a system that upstreams have never really been exposed to, contrary to Linux FreeBSD. percentage up to date, I don't really know why it is still at 96%. To my knowledge there are not so many regressions in the past few years. I guess this accounts for all packages since ever. I have asked for the removal of a hundred outdated packages (Bug#704477), which amounts for 1% only, a fair amount of which are actually just hit by known general FTBFS, or by lack of updated libtool. and (as far as I know) a working port and installer for a modern desktop machine? The installer itself is not a problem. Yesterday I have pushed the few changes pending during the freeze. We have not worked too much on the hardware support in the past months, so it is basically network board drivers from linux 2.6.32, and IDE disk support. I for instance installed it on my Dell D430, and network just works fine. Working on a SATA driver should not be a problem. I just haven't put it high on my TODO list, and have rather worked on the Wheezy release whenever I had time to. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130506202754.gm6...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
Barry deFreese, le Mon 06 May 2013 13:32:04 -0400, a écrit : Fair enough and no offense taken. As I said, I haven't been contributing for quite a while now so they were more questions for myself than anything. Niels actually showed me the graph and it is a bit sadder than I had hoped. What do you mean? I don't see where how it is sad. Nothing happend during the freeze, sure. That's completely expected. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130506202842.gn6...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
Joerg Jaspert, le Mon 06 May 2013 21:27:46 +0200, a écrit : Adding it and then keeping it out of the usual migration rules is asking for failure from the beginning, accumulating cruft. Not a way to go, IMO. Excluding from the migration rules would probably become a headache, yes. But wouldn't just dropping packages which may come in the way of migrations help? From the view of maintainers I think that would be the deciding factor, because it could imply extra work. Not everyone sees the benefits of porting efforts (whereas I see it as excellent QA and promotes better software design, hence I'm in favour of inclusion). I would be in favour of including it if it actually would look like it could be as up to the task as all the rest of the architectures are. But it doesn't appear to me that it is that. Well as up to the task as all the rest is for sure not true. But one can also say the same for kfreebsd or less-supported linux archs. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130506203236.go6...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
Steven Chamberlain, le Mon 06 May 2013 21:20:57 +0100, a écrit : On 06/05/13 20:27, Joerg Jaspert wrote: Adding it and then keeping it out of the usual migration rules is asking for failure from the beginning, accumulating cruft. Not a way to go, IMO. In that case would there be 150-200 RC-severity bugs introduced right away by its inclusion? I would rather say simply dropping them, as already requested in Bug#704477. And as I said a fair amount of these are actually already submitted as general FTBFS bugs or upgrade libtool bugs. There is a lot of overlap though, e.g. fixing a kfreebsd build failure may fix hurd-i386. Yes, there are a lot of these. Funniest cases are for instance that's not linux? Oh, that must be windows, let's include windows.h... Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130506203545.ga11...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
Hi Samuel, On 06/05/13 21:35, Samuel Thibault wrote: Steven Chamberlain, le Mon 06 May 2013 21:20:57 +0100, a écrit : In that case would there be 150-200 RC-severity bugs introduced right away by its inclusion? I would rather say simply dropping them, as already requested in Bug#704477. And as I said a fair amount of these are actually already submitted as general FTBFS bugs or upgrade libtool bugs. If it's possible, yes outdated versions could be removed... and then look again at those figures. But it would need to happen pretty soon. Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/518816c4.7010...@pyro.eu.org
Re: Hurd and the archive
Steven Chamberlain, le Mon 06 May 2013 21:47:00 +0100, a écrit : On 06/05/13 21:35, Samuel Thibault wrote: Steven Chamberlain, le Mon 06 May 2013 21:20:57 +0100, a écrit : In that case would there be 150-200 RC-severity bugs introduced right away by its inclusion? I would rather say simply dropping them, as already requested in Bug#704477. And as I said a fair amount of these are actually already submitted as general FTBFS bugs or upgrade libtool bugs. If it's possible, yes outdated versions could be removed... and then look again at those figures. But it would need to happen pretty soon. Well, I had hoped that it would have happened a month ago already, actually :) Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130506205542.gs6...@type.youpi.perso.aquilenet.fr
Re: Hurd and the archive
Well maybe that was a poor choice of words. I don't even quite understand what some of the columns mean, to be honest. But a visceral response to all of the red and yellow blocks was kind of sad knowing how much you all (and myself to a MUCH lesser degree) have put into it over the last couple of years. I guess I just mean that we seem so close... I didn't mean it in any way to be a slight to you all or the GNU/Hurd in general.. Barry On 5/6/2013 4:28 PM, Samuel Thibault wrote: Barry deFreese, le Mon 06 May 2013 13:32:04 -0400, a écrit : Fair enough and no offense taken. As I said, I haven't been contributing for quite a while now so they were more questions for myself than anything. Niels actually showed me the graph and it is a bit sadder than I had hoped. What do you mean? I don't see where how it is sad. Nothing happend during the freeze, sure. That's completely expected. Samuel -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/51881678.8000...@verizon.net