Re: Using Debian funds to support a gcc development task
Hi! On 9/28/19 11:44 AM, John Paul Adrian Glaubitz wrote: > I'm posting here on behalf the Debian Ports team as we're seeking support > to finance an important development task in gcc. In particular, I'm talking > about the one-time job to modernize the m68k backend by porting it from "CC0" > to "MODE_CC" as described in [1]. > (...) Just as a heads-up: The campaign has been successful. We have collected over $6000 in funds and an experienced GCC developer picked up the task and converted the m68k backend to MODE_CC [1]. I have created an identical campaign for the AVR backend now [2] as this backend is also still using the cc0 representation and some people asked me whether I could create one campaign for AVR as well. Thanks, Adrian PS: I'm not subscribed to the list, so please keep me CC'ed. > [1] > https://www.bountysource.com/issues/80706251-m68k-convert-the-backend-to-mode_cc-so-it-can-be-kept-in-future-releases > [2] > https://www.bountysource.com/issues/84630749-avr-convert-the-backend-to-mode_cc-so-it-can-be-kept-in-future-releases -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
Hi Richard! On 9/29/19 11:26 PM, Richard Z wrote: > On Sat, Sep 28, 2019 at 11:44:26AM +0200, John Paul Adrian Glaubitz wrote: > >> As m68k is the oldest port of both the Linux kernel and Debian after i386, >> it would be a shame to see it go as there is still very good upstream support >> in the Linux kernel with new drivers being added regularly [4], even to >> Linus' >> surprise. > > > in the first place.. wasn't m68k the first ever impelmentation of gcc? Would > be > a shame if they wanted to drop it. Feel free to make a donation to our BountySource campaign if you would like to help [1]. Adrian > [1] > https://www.bountysource.com/issues/80706251-m68k-convert-the-backend-to-mode_cc-so-it-can-be-kept-in-future-releases -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/30/19 4:53 AM, Paul Wise wrote: > On Mon, Sep 30, 2019 at 1:14 AM John Paul Adrian Glaubitz wrote: > >> Having said that, I'm going to unsubscribe from this list now as I realize >> that what we do in Debian Ports is still not considered useful many others, >> so I think it's not possible to find an agreement. > > To be clear, I think the work that folks are doing on the unofficial > Debian ports is valuable and important and that the m68k GCC task is a > good idea. I only dislike using Debian funds to pay people for their > time. I think that crowdfunding the m68k GCC task could work. I think you misunderstood. This is not for paying myself or any other person involved in Debian Ports. This is for collecting money to be able to convince someone experienced in hacking on the gcc code so that they can help us save the m68k backend in gcc. For what is worth, Richard Biener from gcc upstream told me that an experienced gcc developer would be able to perform the task within one months. It's mainly a matter of using enough elbow grease but not necessarily hard work, he says. Since I can't expect a random gcc developer to sink one month of work into working on a backend they are most likely not going to use them- selves, I have started a campaign on BountySource so that I can pay them for the work. I'm not earning a single Cent from this myself. Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/30/19 4:59 PM, Sam Hartman wrote: > Had I been asked for m68k hardware in this instance, I don't think I > would have even blinked before approving the request. The only reason I am asking is because we're currently in a situation where we need external help. We are trying to be as cost-effective as possible and I myself have paid for a lot of things myself, that includes hardware and hosting them. However, in order to solve this particular problem, we need a little more money to succeed which is why I was asking for a single-time donation. Either way, we have already collected 26 donations totaling 1800 US-Dollars now [1], so I'm confident that with a little more help from the community, we'll be able to save the m68k backend in gcc. Regards, Adrian > [1] > https://www.bountysource.com/issues/80706251-m68k-convert-the-backend-to-mode_cc-so-it-can-be-kept-in-future-releases -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
Hi Ian!
Apologies for my late reply.
On 9/30/19 12:40 PM, Ian Jackson wrote:
> Paul Wise writes ("Re: Using Debian funds to support a gcc development task"):
>> To be clear, I think the work that folks are doing on the unofficial
>> Debian ports is valuable and important and that the m68k GCC task is a
>> good idea. I only dislike using Debian funds to pay people for their
>> time. I think that crowdfunding the m68k GCC task could work.
>
> I absolutely agree with this.
>
> John, please let us know if you (or someone else) tries to do this
> via crowdfunding. I promise to contribute.
We have set up a campaign on BountySource, feel free to contribute whatever
amount you want:
> https://www.bountysource.com/issues/80706251-m68k-convert-the-backend-to-mode_cc-so-it-can-be-kept-in-future-releases
Thanks for your kind words and support! Much appreciated.
Adrian
--
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer - glaub...@debian.org
`. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de
`-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
Hello Brock! On 9/29/19 4:10 PM, Brock Wittrock wrote: > 1) It was a simple enough request and reasonable in my opinion. I'm also glad > that he was willing to ask in the first place because as some say, when you > don't ask the answer is already no anyways, so why not ask? > > 2) I understand though why the other side sees his request as a bit > outrageous from a user base and upstream support, perspective (among other > reasons). In most cases, I'd agree that those are the perfect metrics to > measure one's decision on for these kinds of matters in almost all cases. On > the other hand, I see M68K (both the new and old hardware) as an important > architecture to keep around for 1) keeping a wide variety of CPU > architectures available for learning, understanding, and diversity and 2) > historical purposes (although this is obviously the much weaker argument from > a developer support standpoint -- I'd totally agree with that). I'm also sad > to hear about the fate of mips(eb). :( > > 3) It's clear everyone in this thread is passionate about Debian, free > software, and in Adrian's case, passionate about keeping the unofficial M68K > port alive. This passion from everyone is certainly contagious. So kudos to > all of you. > > I hope Debian will reconsider providing at least a small part of the funding > and I'm positive the hobbyist community around M68K (as well as other > avenues) can come together for the rest. I believe by doing so, it would show > Debian doing things that prove itself as "the universal operating system." That was a very kind message. Thanks a lot for confirming I'm not completely alone in this thread. Having said that, I'm going to unsubscribe from this list now as I realize that what we do in Debian Ports is still not considered useful many others, so I think it's not possible to find an agreement. Thanks, Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/29/19 1:17 PM, Sam Hartman wrote: > I'm a bit concerned about your argumentation style in this thread. It > feels to me a lot like you're saying that people are wrong simply > because they are disagreeing with you. In future discussions, I'd > recommend finding a way of having the discussion that acknowledges > disagreement and is more focused on understanding positions than judging > them or persuading others. I don't think that this is the case. I'm just arguing that we have double standards here. You can't say one the one hand, that Debian money should not use for non-Debian work and then spend it on outreachy and you cannot say that architecture must be widely available for use and just emulators and then allow s390x to be a release architecture. > Regardless, I think you have your answer. > Absent the appearance of significant new support, there is not > sufficient interest in spending Debian funds on m68k gcc development. I don't think we have heard enough voices yet to be able to answer that question. Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/29/19 12:48 PM, Hector Oron wrote: > Missatge de John Paul Adrian Glaubitz > del dia dg., 29 de set. 2019 a les 12:20: > >> I communicated the issue to multiple mailing lists, so I assume the active >> m68k people have received my mail. But I haven't heard back from them. > > In particular, have you reached to Andreas Schwab? > (> Is there kernel and toolchain support? At what level? > > Yes. Andreas Schwab does glibc and gcc ) He's a colleague of mine at SUSE, so, yes, I asked him. It's not easy to reach him from time to time so I haven't heard back from him yet (I'm not working in Nuremberg, so I can't just visit his office). >>> Is there any real m68k hardware available nowadays or is it all >>> virtual&emulated? >> >> Yes, there are the Apollo accelerators for one: >> >>> https://www.apollo-accelerators.com/ >> >> And there are embedded open source Coldfire boards being developed: >> >>> http://sysam.it/cff_amcore.html >>> http://sysam.it/cff_stmark2.html > > That looks like MMU-less hardware, would those devices benefit from > running Debian? I assume they are better suited for uClinux > distribution. They use gcc as well. And that doesn't mean there won't be any hardware with MMUs in the near future. At least for Apollo, an MMU is in the make. >> What are the use cases for the port? >> >> The m68k CPU has still a very large hobbyist scene as it's a CPU with an easy >> to learn assembler. There are many hobbyist forums like a1k.org, amiga.org, >> eab.abime.net, www.atari-forum.com and so on with people working on open >> hard- and software projects around the m68k CPU. >> >> As Linus stated in his mail, the Amiga never dies: >> >>> https://lkml.org/lkml/2019/9/24/993 > > Reaching those forums or even organizing a crowd funding campaign > might be more appropriate. That's what I am already doing. I'm not asking Debian alone, of course. >> But I think the list on the page archive criteria is a bit dishonest as >> well when it asks "Are machines available to buy for the general public?" >> while I don't think an IBM Z mainframe is available to buy for the general >> public. >> >> I don't think anyone outside banks, goverments and very large companies >> can buy an IBM Z mainframe, yet Debian provides an official port for these. > > The page is not dishonest, it asks for information, to be able to take > decisions. > In particular for s390x you can read on > https://wiki.debian.org/ArchiveQualification/s390x > It is really up to the teams if they take the burden of the work. If you read that page, you see that a lot of the things mentioned there apply for m68k as well. It even mentions that people are using s390x for just playing around with the architecture. And the argument of emulation is being brought up as well. So, effectively, the s390x port is not more useful to the average Debian user as m68k. > In other words, having hardware available is always great, however > there are special cases and teams decide to accomodate if they believe > it is worth it (at least that's my understanding). But then this should apply for all ports and not just s390x. Thanks, Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/29/19 12:13 PM, Hector Oron wrote: >> Not sure what the problem with LTS is. I thought companies pay for the >> extra effort. I think it's a perfectly fine business model. > > As a very simple summary, companies pay another company (Debian > unrelated) to use Debian volunteers time and Debian resources. > Debian does not get any share of that work. If the work is being paid, it's not volunteers time. That doesn't make sense. You could argue about Debian resources being used without paying Debian. Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/29/19 12:07 PM, Hector Oron wrote: >> So, would -project be willing to support our cause through Debian funds? > > I have been looking at > https://wiki.debian.org/ArchiveQualification/m68k, I was expecting > some answers to the questions proposed at > https://ftp-master.debian.org/archive-criteria.html, it looks to me > that ArchiveQualification/m68k wiki page needs updating. > > Have you also checked with m68k porters listed in that page? (I assume > most of them are retired from m68k work) I communicated the issue to multiple mailing lists, so I assume the active m68k people have received my mail. But I haven't heard back from them. > Is there any real m68k hardware available nowadays or is it all > virtual&emulated? Yes, there are the Apollo accelerators for one: > https://www.apollo-accelerators.com/ And there are embedded open source Coldfire boards being developed: > http://sysam.it/cff_amcore.html > http://sysam.it/cff_stmark2.html What are the use cases for the port? The m68k CPU has still a very large hobbyist scene as it's a CPU with an easy to learn assembler. There are many hobbyist forums like a1k.org, amiga.org, eab.abime.net, www.atari-forum.com and so on with people working on open hard- and software projects around the m68k CPU. As Linus stated in his mail, the Amiga never dies: > https://lkml.org/lkml/2019/9/24/993 But I think the list on the page archive criteria is a bit dishonest as well when it asks "Are machines available to buy for the general public?" while I don't think an IBM Z mainframe is available to buy for the general public. I don't think anyone outside banks, goverments and very large companies can buy an IBM Z mainframe, yet Debian provides an official port for these. Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/28/19 9:07 PM, Guillem Jover wrote: > I'm afraid this argument cuts both ways. I would find it extremely > demotivating if Debian started spending money to pay people to work > on tasks that up to now have been volunteer based (where volunteer > of course can include a company volunteering employees time f.ex.). I don't think any of the projects like gcc, the kernel, binutils or glibc are maintained by volunteers these days. At least I don't know any volunteer doing that work on a larger scale. At SUSE, we have dedicated teams who work on gcc, the kernel, kvm and so on. I don't think anyone - unless they are rather wealthy - can afford to spend working all day on a project without getting paid. > I already find the LTS effort borderline, where I try to refuse out > of principle to do LTS work for packages I maintain, and where I've > found myself being pretty unhappy that one time I had to do some of > that for a native package, which would have had a very weird release > history otherwise. :/ Not sure what the problem with LTS is. I thought companies pay for the extra effort. I think it's a perfectly fine business model. Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
On 9/29/19 3:14 AM, Paul Wise wrote: > On Sat, Sep 28, 2019 at 5:44 PM John Paul Adrian Glaubitz wrote: > >> Since Debian is also supporting projects for a good cause using their funds, > > Do you have any examples of this? AFAIK we don't support development > nor external projects using Debian funds. The only exception I can > think of is helping Outreachy interns join the FLOSS community through > contributing to Debian related projects. You just gave one example yourself. >> We already have a potential developer > > I think it inappropriate to select the developers to reward with money > in this way. Instead (once funded) the job should be posted to GCC & > m68k related forums, FOSSJobs and other FLOSS paid work aggregators > and the best candidate selected. I didn't say that we have already chosen one person to do the job. Just that there is already one potentially interested of doing it. I don't have any particular preference in which person is supposed to do it. >> So, would -project be willing to support our cause through Debian funds? > > Traditionally, Debian has not used our funds to support development > work, the only exceptions being Outreachy and Dunc-Tank. I think that > we should keep it that way, no matter what the project is. The work that we have done in Debian Ports has positively affected a lot of other parts of Debian. We've done a lot of clean up in debian-installer, made Rust available on all release architectures, helped fixing tons of bugs in many Debian packages, helped Helmut with rebootstrap and made it easier to bootstrap Debian on a new architecture like riscv64. I don't think a one-time donation from the project is too much to ask for it in return. > Debian is just a small part of the FLOSS community, I think a better > way to fund FLOSS development that companies don't want to touch would > be a community-wide non-profit organisation (such as "The Free > Software Endowment Inc.") focussed on this. Another option is > crowdfunding individual projects; this has been used successfully for > development (Bootlin's work on Allwinner hardware for example) so it > could work for this situation too. I actually remembered what Neil, our DPL in 2016, reported during his talk at DebConf16. He asked the project to spend Debian money in 2015 because it was accumulating too much. Then more money was spent and as a result, the project had more money in 2016 than it had in 2015. Adrian -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: Using Debian funds to support a gcc development task
> On Sep 28, 2019, at 7:57 PM, Tollef Fog Heen wrote: > > ]] John Paul Adrian Glaubitz > >> I don’t know what “m-f-t” stands for in this context, sorry. I’m on >> mobile at the moment though so my phone might be messing up >> things. Sorry for that. > > Mail-Followup-To. Don't Cc people unless explicitly requested. > > [...] I see. But normally any reasonable mail client is able to cope with that. At least mutt and Thunderbird do. But I have removed that CC now. >> But that’s just your personal opinion on what the focus should be on >> when supporting a good cause. Someone else could argue about more >> diversity in software. We have something as the “Debian init >> diversity” project after all which is also a non-commercial but a >> community effort. > > I don't get your focus on commercial vs non-commercial here, I think > you're the only person in the thread talking about commercial concerns > as something that even enters the picture. It’s very simple. If there wasn’t any commercial interest in Linux on $ARCH, it would have stopped being a release architecture in Debian long time ago. Or do you actually know someone who has got a IBM mainframe at home (s390x)? All architectures in Debian that are release architectures are ones that are commercially supported. So if you argue Debian should support release architectures only, you are implicitly arguing that Debian funds should not be used for any targets that have no commercial relevance. >> I think it’s up to every free software developer which cause they >> would like to support. After all, free software also means we work on >> the projects we are passionate about and not what’s commercially >> viable. > > Sure; feel free to support the m68k porting effort as much as you want > and in any reasonable fashion you want. Nobody is going to stop you. > > I'm arguing against spending Debian money on toolchain maintenance (for > a port that's no longer part of Debian proper even!). Not what you or > GCC upstream or anybody else does with their own time and money. But my point is that everyone has a different focus on what good cause they would like to support and I think we agree that commercial viability the only criteria. Adrian
Re: Using Debian funds to support a gcc development task
> On Sep 28, 2019, at 7:19 PM, Tollef Fog Heen wrote: > > > Please respect m-f-t, as is the custom on Debian lists? I don’t know what “m-f-t” stands for in this context, sorry. I’m on mobile at the moment though so my phone might be messing up things. Sorry for that. > ]] John Paul Adrian Glaubitz > >> As I explained in my previous mail: The development task here is >> something that goes a little beyond normal maintenance work and hence >> requires someone to work with a longer dedication on the task. > > The required level of maintenance varies over time, that's completely > normal, and I don't see how this changes anything. Well, it would mean that software projects are not commercially supported are not in Debian’s interest. >> While gcc is free software, it doesn’t mean the work on it is free. I >> think we all know that without commercial support, free software >> wouldn’t be able to survive these days. > > Of course not; everybody needs to put food on the table, one way or the > other. Some of us are paid to work on Debian and free software and do > it that way. Some do it during our free time, either because they earn > enough that they can do it as a hobby or because they are a student with > free time on their hands, or some other reason that makes it possible > for them to contribute without getting paid for it. This hasn't really > changed in a very long time. Okay, so we agree on this part. Some work requires paid developers due to the extensive work required. If you look at the MAINTAINERS files for gcc, the kernel or other relevant projects, you’ll see that they are often maintained by large companies like IBM. But that doesn’t mean Debian should be focused on projects of commercial interest only, does it? >>> Keeping the toolchain working is a pretty essential requirement for >>> keeping a port alive, and I don't think it's viable to base the ongoing >>> toolchain maintenance for a port on fundraising. >> >> Maintenance isn’t the same as a one-time porting effort. Normal target >> maintenance work is usually a matter of discovering bugs and fixing >> them unless you are a port with commercial support where paid >> developers are working on supporting new features and hardware on a >> regular basis. > > Maintenance effort over time by far exceed the initial porting cost, so > if the port isn't even able to surmount that, I don't think it's > long-term viable. I don’t think this is something that can be generally applied. The fact that BountySource and GSOC exist are the very proof that it’s perfectly normal to support one-time development tasks through funding efforts. If free software would only be about commercial interests, nothing but SLES and RHEL on x86_64 and POWER would probably exist. > [...] > >>> As a general rule, I don't think Debian should pay developers to write >>> software. (There are some exceptions such as outreachy, but they are >>> few.) >> >> Does that mean you would agree to supporting the effort if the >> developer came from a minority group? (It might actually be the case >> here.) > > No, it means that there are situations where I think giving people from > less-privileged backgrounds a leg up so they can start contributing > might be appropriate. The suggested project does not sound like a > project for somebody who is not already contributing to GCC. I guess > you could try to do it as a GSoC project if it's in that ballpark. But that’s just your personal opinion on what the focus should be on when supporting a good cause. Someone else could argue about more diversity in software. We have something as the “Debian init diversity” project after all which is also a non-commercial but a community effort. After all, everyone has different ideas in which regards they would like to support free software projects and that’s perfectly fine. For some people, it’s supporting less represented groups among developers, others support less common init systems and others like Debian Ports support less popular architectures. I think it’s up to every free software developer which cause they would like to support. After all, free software also means we work on the projects we are passionate about and not what’s commercially viable. Thanks, Adrian
Re: Using Debian funds to support a gcc development task
> On Sep 28, 2019, at 6:19 PM, Tollef Fog Heen wrote: >> >> No, it just means that the current gcc maintainer [1] for m68k backend hasn't >> worked on this particular task yet because his employer wouldn't pay for >> this particular work. Unlike the other ports like amd64, ppc64el, arm* >> and s390x, we don't have large companies supporting us as the commercial >> potential is low although there is still a small Amiga, Atari and Mac68k >> market with new hardware and software being made. >> >> gcc is just one part of the port, others parts like the Linux kernel or >> the Debian ports are actively maintained as I mentioned in my initial mail. > > To me your «no» actually means «yes». When we're talking manpower, it's > about the right people with available time and ability. It's not about > the number of warm bodies, so if there's just a single person who is > able to do this work and they don't have the time, the port is missing > absolutely critical manpower. As I explained in my previous mail: The development task here is something that goes a little beyond normal maintenance work and hence requires someone to work with a longer dedication on the task. While gcc is free software, it doesn’t mean the work on it is free. I think we all know that without commercial support, free software wouldn’t be able to survive these days. All I am asking is for a one-time donation. > Keeping the toolchain working is a pretty essential requirement for > keeping a port alive, and I don't think it's viable to base the ongoing > toolchain maintenance for a port on fundraising. Maintenance isn’t the same as a one-time porting effort. Normal target maintenance work is usually a matter of discovering bugs and fixing them unless you are a port with commercial support where paid developers are working on supporting new features and hardware on a regular basis. >>> I don't think spending $1-5k would be the best use of Debian >>> funds. >> >> Is that really that amount of money? Paying a developer is normally >> a lot more expensive. > > You were the one who suggested that sum, not me. > > As a general rule, I don't think Debian should pay developers to write > software. (There are some exceptions such as outreachy, but they are > few.) Does that mean you would agree to supporting the effort if the developer came from a minority group? (It might actually be the case here.) Kind Regards, Adrian
Re: Using Debian funds to support a gcc development task
Hello! On 9/28/19 3:26 PM, Tollef Fog Heen wrote: >> Since the lack of modernization would eventually mean that m68k support would >> get removed from gcc, I'm currently running a campaign to prevent that. I >> have already opened a tracker bug upstream in gcc's bugzilla [2] as well as >> linked the issue to BountySource [3]. > > Doesn't this just mean there's not enough manpower to keep the port > alive? No, it just means that the current gcc maintainer [1] for m68k backend hasn't worked on this particular task yet because his employer wouldn't pay for this particular work. Unlike the other ports like amd64, ppc64el, arm* and s390x, we don't have large companies supporting us as the commercial potential is low although there is still a small Amiga, Atari and Mac68k market with new hardware and software being made. gcc is just one part of the port, others parts like the Linux kernel or the Debian ports are actively maintained as I mentioned in my initial mail. If I had the necessary gcc experience to work on this, I would do it myself. But at the moment, we have to rely on external help. > I don't think spending $1-5k would be the best use of Debian > funds. Is that really that amount of money? Paying a developer is normally a lot more expensive. > As you point out, it's one of the oldest ports, but ports go through a > natural life cycle where they eventually pass away, and that's ok. There is a very active community around the port so there are people using it, although it's not for commercial purposes, of course. Thanks, Adrian > [1] https://github.com/gcc-mirror/gcc/blob/master/MAINTAINERS#L80 -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Using Debian funds to support a gcc development task
Hello! I'm posting here on behalf the Debian Ports team as we're seeking support to finance an important development task in gcc. In particular, I'm talking about the one-time job to modernize the m68k backend by porting it from "CC0" to "MODE_CC" as described in [1]. In the future, gcc upstream expects all backends to be using MODE_CC for the internal register representation as the old CC0 is supposed to be removed. Since the lack of modernization would eventually mean that m68k support would get removed from gcc, I'm currently running a campaign to prevent that. I have already opened a tracker bug upstream in gcc's bugzilla [2] as well as linked the issue to BountySource [3]. As m68k is the oldest port of both the Linux kernel and Debian after i386, it would be a shame to see it go as there is still very good upstream support in the Linux kernel with new drivers being added regularly [4], even to Linus' surprise. I have already gained multiple supporters on the largest German Amiga forum a1k.org as well some people through the Debian/m68k mailing list. We haven't started with the funding itself yet since we still need to coordinate some things first, the primary decision being which funding platform we are going to use. Since Debian is also supporting projects for a good cause using their funds, I was wondering whether the project would also be willing to help support our cause to modernize the m68k backend in Debian with a single donation, I thought of something around $1000 to $5000 depending on how much the project is willing to spend. We already have a potential developer to work on the task who has a lot of experience with gcc and of whom I'm confident he can solve this task before gcc-11 is going to be released. I have already talked to the DPL personally and he recommended me to ask on debian-project to receive feedback from a broader audience of the Debian project. So, would -project be willing to support our cause through Debian funds? Thanks, Adrian > [1] https://gcc.gnu.org/wiki/CC0Transition > [2] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91851 > [3] > https://www.bountysource.com/issues/80706251-m68k-convert-the-backend-to-mode_cc-so-it-can-be-kept-in-future-releases > [4] https://lkml.org/lkml/2019/9/24/993 -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Re: [Debian-ports-devel] [lfili...@debian.org: scheduled downtime for Debian services at UBC (2016-01-09T16:00Z/20:00Z)]
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Hi Luca! Is there a way to contact staff at UBC who admin the network infrastructure at UBC? We are seeing lots of timeouts for packages being upload to leda.debian.net, independent of the location and network of the buildd. We have ports buildds located in the US, Hungary and Germany and they are all getting timeouts when running dupload. There was no configuration change on the sides of the buildds and since the machines are located in completely different networks, I suspect it must be an issue at UBC. Maybe the switch that leda is connected to is dropping packages or someone on site connected a loop on the switches. In any case, most uploads are timing out which is becoming more and more a problem. If you have any idea what we could do or whom we could contact, I'd highly appreciate your input. Cheers, Adrian On 01/09/2016 02:12 AM, Aurelien Jarno wrote: > This affects leda, see below. > > - Forwarded message from Luca Filipozzi - - > > From: Luca Filipozzi > To: debian-infrastructure-annou...@lists.debian.org > User-Agent: Mutt/1.5.23 (2014-03-12) > X-Spam-Status: No, score=-20.1 required=4.0 tests=DKIM_SIGNED,DKIM_VAL ID, > DKIM_VALID_AU,DKIM_VERIFIED,FOURLA,HEADER_FROM_DIFFERENT_DOMAINS, > LDO_WHITELIST,MURPHY_DEBIAN_MESSAGE,PGPSIGNATURE,RP_MATCHES_RCVD auto learn=unavailable autolearn_force=no version=3.4.0 > X-Spam-Status: No, hits=-4.7 > Date: Sat, 9 Jan 2016 00:56:00 + > Subject: scheduled downtime for Debian services at UBC (2016-01-09T16: 00Z/20:00Z) > Mail-Followup-To: debian-project@lists.debian.org > > Colleagues, > > DSA was advised a few moments ago that there is a scheduled power outa ge at UBC > that will impact Debian services. Specifically, power to the MCLD bui lding > will be disrupted, impacting Debian machines in the MCLD server room d irectly > and in the KAIS server room indirectly. > > The following hosts are in the MCLD server room (power outage): > > * spontini.debian.org > > The following hosts are in the KAIS server room (network outage): > > * blavet.debian.org > * buxtehude.debian.org > * danzi.debian.org > * diabelli.debian.org > * elgar.debian.org > * fano.debian.org > * finzi.debian.org > * geo2.debian.org > * glinka.debian.org > * gombert.debian.org > * jenko.debian.org > # leda.debian.net > * lotti.debian.org > * lucatelli.debian.org > * menotti.debian.org > * muffat.debian.org > * nono.debian.org > * reger.debian.org > * sonntag.debian.org > * tchaikovsky.debian.org > * traetta.debian.org > * tye.debian.org > * ubc-bl2.debian.org > * ubc-bl3.debian.org > * ubc-bl4.debian.org > * ubc-bl6.debian.org > * ubc-bl7.debian.org > * ubc-bl8.debian.org > * ullmann.debian.org > > Thanks, > > Luca Filipozzi > Debian System Administration > > > > _______ > Debian-ports-devel mailing list > debian-ports-de...@lists.alioth.debian.org > https://lists.alioth.debian.org/mailman/listinfo/debian-ports-devel > - -- .''`. John Paul Adrian Glaubitz : :' : Debian Developer - glaub...@debian.org `. `' Freie Universitaet Berlin - glaub...@physik.fu-berlin.de `-GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913 -BEGIN PGP SIGNATURE- Version: GnuPG v2 iQIcBAEBCAAGBQJWklJIAAoJEHQmOzf1tfkTTT8P/28342XllbRkgjy81w2+QhO0 cT66bS4o1oINRReDxaOmfYV91kz7mT12GNa/XIxXGvNGhnSnyDk1nHfjMwrUg+NN iVEIv4bWh71p+v4XoQ0uxDCVuEr5EkBjmvc2EihLqmuAuQACqu2C/seU1uy+P4Hi B2B0o059kbSvs6lfoXrwTyFEHG7jiajMOVSsaC/ht981/6S+z/2YIOkTC3FND0KK 3RiFYSoxMfiTLAc9AZKYk61dQUSC7qYKwD+G1B7HeJzLC7eyER3pm0rwxmBrZZs2 d1uDVqSKfFcPLhBn2M292XHIkZIU+uxOl9f4zMnuk+62SMc9zKEacZLudwMdGWKX FSy0iFn1YnCeqg02ED09I8QKNlhmAN7wf4mIBZkct9rgPZnT+HIgmovTeRtAZIkj 11eFr27N7B0HpKEaMNR8x1VJyPT2gbm0F3bQw+Lx9rD5m0gf6jn5OYIYMlN+Bc+o xVP0ZQKHC9rH9C+XAH/Pn9eUJ6PLABzbmoerhbDIo7G+1FWhrwnhJ8gC1S9vXy/T nZIvc3L3s1cydwiW7eRuGY+8iSjZNM7uMyzyjJ6PydcfX6rf7cV9gZQ0+4yGpPnp cLLovBodcG/Vrz1lW2MkWmWV1NSdwraDr5Uhuvs/GCKo2IXDErXijhk0yo2rabug qfpSRljLAMs3NYpLzyqi =tO69 -END PGP SIGNATURE-
