Re: [Sugar-devel] Community XO software builds
On Fri, May 8, 2015 at 8:18 AM, Samuel Greenfeld wrote: > You might laugh but when the OLPC Association was actively answering bids > for laptops, this dance happened all the time. Also known as "sales calls", never a strong point with FLOSS projects, and one of the reasons so many Intel Classmates were deployed. Sean. ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 7, 2015 at 7:49 PM, James Cameron wrote: > On Thu, May 07, 2015 at 11:28:42AM -0300, Gonzalo Odiard wrote: > > When I talked with deployments and they ask for Ubuntu, > > and I ask why, what they really want is Long Time Support. > > No deployment change their image more than once a year. > > In fact, change a image is a logistic challenge for most of > > the big/middle size deployments. > > This continues to puzzle me. LTS is a stream of security updates, and > you say the deployments do not apply them until the next year? > > And yet they want them? > > They want something they don't use? > > If a vulnerability is reported just after they make their image, the > children are exposed to the vulnerability for the rest of the year. > > It seems more likely that the meaning of LTS is not understood. > > Fedora continues with security updates for a similar time period, but > if the deployment uses our builder unchanged they won't get them. I'm > expecting that if a deployment needs LTS on Fedora they will assume > the responsibility to apply the updates when they make a build. > > All valid points. I sent a email to the deployment to ask for more information. I will report when have a reply. -- Gonzalo Odiard SugarLabs - Software for children learning ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Fri, May 8, 2015 at 12:04 AM, James Cameron wrote: > On Thu, May 07, 2015 at 10:28:06AM +0100, Peter Robinson wrote: >> On Thu, May 7, 2015 at 4:21 AM, Samuel Greenfeld >> wrote: >> > The obvious counterargument would be that a deployment might want >> > to deploy your XO-Next (whatever it is) alongside existing XO >> > laptops, allowing all of them to have the same configuration. >> >> From my memory of olpc-os-builder it was very modular and wouldn't be >> hard to add dozens of different devices support to it. > > Yes, it would be straightforward to add commodity hardware support to > olpc-os-builder. Add kernel and boot loader. Add some sort of > installer. > > But we have SoaS, and SoaS works fine on commodity hardware, so why > bother with olpc-os-builder? Because same process for any and every device. A single process is a good thing, it makes it easier to understand and get a consistent configuration everywhere. > Because olpc-update? Nobody uses it. The interesting thing here is that Atomic on Fedora would provide everything that olpc-update was designed to do and it could make upgrades between Fedora releases much easier and less of an issue with regards to TLS. Plus probably a bunch of things that it currently doesn't and it's upstream being actively developed, instead of home brew, would likely ease the security updates issues mentioned previously and easy pushing out of updates, caching updates for bandwidth etc. > Because preinstalled activities? SoaS can do that too. > >> > There's plenty of blame to go around in terms of re-inventing the wheel and >> > lack of communication. >> >> Yep! >> >> > There simply (and correct me if I'm wrong) are not the resources inside of >> > OLPC, outside, or combined at this time to maintain and update two separate >> > builds & build systems. >> > >> > It amazes me how far we bend over backwards to avoid saying "end of life" >> > and "end of support". >> > >> > >> > I have seen a fair amount of interest, both publicly and privately, for >> > newer XO laptop builds. But I don't think the requesters realize how much >> > work it takes to make one. >> >> The big one here is kernel kernel kernel. > > Yes. > >> > And I do not forsee anyone stepping up to get the XO-1.75 and XO-4 kernel & >> > drivers into a state they can be upstreamed or upgraded for newer Fedoras >> > unless a deployment really wants this instead of newer equipment. >> >> Or even the 1.5, I believe most of the XO-1 support is upsteream. >> >> > Newer operating systems tend to require more disk space and RAM than the >> > predecessors. We have seen this even within Fedora's lineage. >> >> Yes, and no. I mean 1Gb of the original XO-1 is tight, but SoaS still >> happily fits in 4Gb with a bunch of space to spare. Looking at my >> current SoaS VM the used space is around 1.9Gb. Amusingly the various >> cloud/container enterprise initiatives actively help us here because >> for once they care about dependency bloat too :-) >> >> The two things that add bloat to the current SoaS image are: >> * Browse needs to be converted to the new WebKitGtk APIs so we don't >> ship two copies of WebKitGtk. >> * Conversion of remaining gstreamer 0.10 to 1.0 to allow us not to ship that. >> >> Ultimately I think you could with a little development effort get it >> down to 1.5Gb used space which would make a 2Gb filesystem quite >> usable. >> >> > Since OLPC already appears to be going the Ubuntu LTS route, I would argue >> > it would be easiest to take everything that way, porting utilities as >> > required, and make that the final image & build system for XOs. >> >> Personally I have no interest in that. I wish you luck. > > -- > James Cameron > http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
There are at least two types of "deployments"/"customers" that Sugar has. The first is the small, volunteer group. To them, it doesn't matter what OS they actually are using, or (to some extent) how well tested things are. They just want to come in, try something with their students, and if they need to tweak something or something breaks, it's no big deal. The second is the large deployment. And large deployments, like large corporations, do not want to deploy Sugar widely unless they have a chance to thoroughly check it out. First, they might investigate a bit to see who currently uses Sugar, and if there are any other users they can get recommendations from. Then they might look into Sugar Labs, asking about Sugar's history, what warranties were available, the future roadmap for features, etc. They may insist on having a face-to-face meeting with a Sugar representative, where they could ask detailed questions. You might laugh but when the OLPC Association was actively answering bids for laptops, this dance happened all the time. When large corporations sell things to each other, support can be everything. It doesn't mean that they are going to use it. But if they need a patch for critical bug on the President's laptop, or the latest Shellshock or Heartbleed that their bosses' boss' saw in the news, they want to have something or someone they can point to definitely get support. Very few deployments have invested in the resources to internally make their own OS images at that level of detail. I don't want to go into it too much in this email, but dealing with large organizations can be a very different thing. On Fri, May 8, 2015 at 1:37 AM, Tony Anderson wrote: > I don't know what is puzzling. I can understand a deployment wanting > assurance of long-term support for Sugar. I doubt there are many > deployments that even know what Fedora or Ubuntu means. Even fewer that > understand the difference between SugarLabs and Red Hat or Canonical as > sources of this support. > > The word deployment may be a puzzle, In some cases it as a national > ministry or OLPC Australia. For most of us, it is a school or other > institution which has acquired OLPC laptops and is attempting to make use > of them. > > There are many deployments which have never updated their image. In > general, an update to an XO requires someone to come to the school > with the technical expertise to do so. I am sure there are schools which > have never seen such a visitor since they received their laptops. > The positive element is that the laptops work as they always have. The > downside, of course, is that the users have no chance to benefit from > the new capabilities available from current releases. > > Finally, what urgent security fixes are required by a deployment with no > access to the internet? > > Tony > > > On 05/08/2015 12:55 AM, sugar-devel-requ...@lists.sugarlabs.org wrote: > >> When I talked with deployments and they ask for Ubuntu, >>> >and I ask why, what they really want is Long Time Support. >>> >No deployment change their image more than once a year. >>> >In fact, change a image is a logistic challenge for most of >>> >the big/middle size deployments.? >>> >> This continues to puzzle me. LTS is a stream of security updates, and >> you say the deployments do not apply them until the next year? >> >> And yet they want them? >> >> They want something they don't use? >> > > ___ > Sugar-devel mailing list > sugar-de...@lists.sugarlabs.org > http://lists.sugarlabs.org/listinfo/sugar-devel > ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 07, 2015 at 10:28:06AM +0100, Peter Robinson wrote: > On Thu, May 7, 2015 at 4:21 AM, Samuel Greenfeld wrote: > > The obvious counterargument would be that a deployment might want > > to deploy your XO-Next (whatever it is) alongside existing XO > > laptops, allowing all of them to have the same configuration. > > From my memory of olpc-os-builder it was very modular and wouldn't be > hard to add dozens of different devices support to it. Yes, it would be straightforward to add commodity hardware support to olpc-os-builder. Add kernel and boot loader. Add some sort of installer. But we have SoaS, and SoaS works fine on commodity hardware, so why bother with olpc-os-builder? Because olpc-update? Nobody uses it. Because preinstalled activities? SoaS can do that too. > > There's plenty of blame to go around in terms of re-inventing the wheel and > > lack of communication. > > Yep! > > > There simply (and correct me if I'm wrong) are not the resources inside of > > OLPC, outside, or combined at this time to maintain and update two separate > > builds & build systems. > > > > It amazes me how far we bend over backwards to avoid saying "end of life" > > and "end of support". > > > > > > I have seen a fair amount of interest, both publicly and privately, for > > newer XO laptop builds. But I don't think the requesters realize how much > > work it takes to make one. > > The big one here is kernel kernel kernel. Yes. > > And I do not forsee anyone stepping up to get the XO-1.75 and XO-4 kernel & > > drivers into a state they can be upstreamed or upgraded for newer Fedoras > > unless a deployment really wants this instead of newer equipment. > > Or even the 1.5, I believe most of the XO-1 support is upsteream. > > > Newer operating systems tend to require more disk space and RAM than the > > predecessors. We have seen this even within Fedora's lineage. > > Yes, and no. I mean 1Gb of the original XO-1 is tight, but SoaS still > happily fits in 4Gb with a bunch of space to spare. Looking at my > current SoaS VM the used space is around 1.9Gb. Amusingly the various > cloud/container enterprise initiatives actively help us here because > for once they care about dependency bloat too :-) > > The two things that add bloat to the current SoaS image are: > * Browse needs to be converted to the new WebKitGtk APIs so we don't > ship two copies of WebKitGtk. > * Conversion of remaining gstreamer 0.10 to 1.0 to allow us not to ship that. > > Ultimately I think you could with a little development effort get it > down to 1.5Gb used space which would make a 2Gb filesystem quite > usable. > > > Since OLPC already appears to be going the Ubuntu LTS route, I would argue > > it would be easiest to take everything that way, porting utilities as > > required, and make that the final image & build system for XOs. > > Personally I have no interest in that. I wish you luck. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 07, 2015 at 03:59:10PM +0100, Peter Robinson wrote: > Has anyone actually booted the latest Ubuntu LTS on any/all the XOs? I've tried it on XO-1.5, and it was doable, but lots of missing things. My estimate to fix is greater than the size of work to stabilise Fedora 20 or Fedora 22. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 07, 2015 at 11:28:42AM -0300, Gonzalo Odiard wrote: > When I talked with deployments and they ask for Ubuntu, > and I ask why, what they really want is Long Time Support. > No deployment change their image more than once a year. > In fact, change a image is a logistic challenge for most of > the big/middle size deployments. This continues to puzzle me. LTS is a stream of security updates, and you say the deployments do not apply them until the next year? And yet they want them? They want something they don't use? If a vulnerability is reported just after they make their image, the children are exposed to the vulnerability for the rest of the year. It seems more likely that the meaning of LTS is not understood. Fedora continues with security updates for a similar time period, but if the deployment uses our builder unchanged they won't get them. I'm expecting that if a deployment needs LTS on Fedora they will assume the responsibility to apply the updates when they make a build. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 07, 2015 at 10:16:42AM +0100, Peter Robinson wrote: > On Thu, May 7, 2015 at 12:10 AM, James Cameron wrote: > > On Wed, May 06, 2015 at 09:19:45AM -0300, Gonzalo Odiard wrote: > >> I think we should try make a build using CentOS. I don't know if > >> have all the packages we need, but the rate of change in Fedora was > >> difficult to follow when OLPC had a team dedicated and now is almost > >> impossible. The true is we didn't finished to solve the problems we > >> found in F20, and Fedora is working in F22. > > > > I do not think we should switch from Fedora to CentOS, because; > > > > 1. our installed base express interest in Fedora or Ubuntu, > > Daniel Drake, myself and others put in a lot of effort back in the > F-14/15 days to get everything upstream into Fedora. I continue to > maintain that and produce a Sugar on a Stick release with every Fedora > release. Agreed, and my continued thanks. > In the last release Daniel and I was involved in the delta between > Fedora and the OLPC release was very minimal. Basically kernel, > firmware, and some minor changes to a couple of Sugar packages for XO > HW and patches that weren't yet upstream. > > > 2. there are missing desktop packages, which means we are taking on > > maintenance of those packages on CentOS, > > Having tried and failed to do this back when EL6 was new I believe > this is a dead end. It turned out to be _WAY_ more effort than > actually keeping Fedora up to date. The upstream RHEL releases are > every 6 months but if you need a fix for a package in the core 2500 > odd packages and it's not easy you might be waiting a lot longer for a > fix. > > In Fedora if you know the right people (like me) you can get a fix > into update-testing in a day. Also there's a much much wider QA group > across the packages we use and care about. > > I can go on and on about the details required for this but basically I > suspect eyes have glazed over already. > > > 3. we would delay necessary work until the next release of CentOS, or > > if the work is too large we may never upgrade. > > I suspect it would be never. > > > Let me explain that last point. > > > > There is a continuous flow of changes into Fedora. These changes > > eventually flow into Red Hat Enterprise Linux, and thus into CentOS. > > > > The most cost effective way to handle this flow was for developers to > > test changes on our builds, every week. This gaves us awareness of > > the change and kept us involved to resist changes that cause damage. > > We were there once. It required a low but continuous engineering > > effort. > > It use to take around an hour to cut a release from Fedora/Sugar > repos. Quite often the delta from a patch for a fix being created and > a new OS was in the hours timeframe. It's the usual story of a little > bit of effort regularly stops it from being a major issue. > > Kernel and olpc-os-builder aside I think you could probably produce a > working image of Fedora 22 now. I think all the userspace bits are > likely there and working due to my SoaS work. Agreed. I've not tried Fedora 22 yet. Time poor. > It's actually the thing that annoys me most about the sugar community. > IMO we have a great working Sugar release that works pretty much > everywhere plus is a great proven base for XO releases yet so many > core developers have told me "if only you'd focus on Ubuntu we'd use > it" yet Ubuntu for _YEARS_ have shown that they couldn't given a shit > and even actively remove core bits needed (remember the Browse on > Mozilla years anyone??) to make it even harder. Yes. Mostly I think the calls for Ubuntu spring from familiarity with it, and an unwillingness to engage in the process for fixing problems with Fedora (OLPC and SoaS). The Ubuntu LTS doesn't cover Sugar, so it is wrong to expect value from it. > > The next most cost effective way is to do this work only when a new > > release of Fedora occurs. This results in lots of head scratching and > > bug fixing, and new builds, until the bugs are mostly gone. We are > > here now. It requires bursts of engineering effort. > > Actually it needs work _BEFORE_ a new release happens, any work now > IMO should be focused on Fedora 23. That way you have everything in > place in time for Fedora 23 GA in October and you get the longest > value out of the release. > > > The least cost effective way is to hold off doing that work for three > > years until the next CentOS release. This would be a lot more work in > > a much shorter burst. > > And you'll likely end up in a very disparate stability across devices. > Both ARMv7 and i686 is community supported in CentOS which means you > get likely dubious quality of work and I suspect due to toolchain > config choices for i686 it won't even run on the XO-1. Has anyone > actually tried booting CentOS-7 on a XO1? From what I've seen of the > ARMv7 efforts I see it as half arsed at best. No, nobody has tried booting CentOS-7 on an XO-
Re: [Sugar-devel] Community XO software builds
No official announcements yet. Samuel was looking at the bug database, because he knows where to look :) Gonzalo On Thu, May 7, 2015 at 7:45 AM, Martin Dengler wrote: > On Wed, May 06, 2015 at 11:21:35PM -0400, Samuel Greenfeld wrote: > > OLPC already appears to be going the Ubuntu LTS route > > When/where can I read more about what OLPC is doing with Ubuntu LTS? > Apologies > for the lazyweb request. > > Martin > > ___ > Devel mailing list > Devel@lists.laptop.org > http://lists.laptop.org/listinfo/devel > > -- Gonzalo Odiard SugarLabs - Software for children learning ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
> Ok. I didn't know that. > > When I talked with deployments and they ask for Ubuntu, > and I ask why, what they really want is Long Time Support. > No deployment change their image more than once a year. > In fact, change a image is a logistic challenge for most of > the big/middle size deployments. Yes, but in the Ubuntu case it's a one way street, there's never been any love from Ubuntu. You could get it from Canonical if someone put up a large amount of cash. They don't support gnome in the core so even then I suspect you'll have similar issues. > Then, I was thinking in CentOS as a LTS version of Fedora. Nope, never has been never will be. It's a _DOWNSTREAM_ of RHEL, and in v7 RHEL only supports 64 bit architectures. The i686 support is community driven as is the ARMv7 support. No guarantee either will ever be in sync. Has anyone actually booted the latest Ubuntu LTS on any/all the XOs? >> In Fedora if you know the right people (like me) you can get a fix >> into update-testing in a day. Also there's a much much wider QA group >> across the packages we use and care about. >> >> I can go on and on about the details required for this but basically I >> suspect eyes have glazed over already. >> > > This is true, and I know that. > But also is true, that keep the pace of changes in Fedora is not easy. > In fact, is not Fedora fault, mostly is Gtk ([1], [2], [3]) or libraries > (the last was vte [4], > but I can find more). Yes, I'm aware of that. There's a number of other bits I'm aware of. Be aware it's not all rosy in the LTS world either. >> > 3. we would delay necessary work until the next release of CentOS, or >> > if the work is too large we may never upgrade. >> >> I suspect it would be never. > > > Ok. But Let me explain my reasons. > > Right now, the only "stable" images are based on F18. > We don't have images in a good shape for the deployments for F20, > we missed F21 (where Gtk theme change in a subtle way again, > and toggle toolbar buttons don't change the background color), > and we should start to work in F22. With the hands we have today, > I am sure we will not solve all the problems we already have before F23 is > released. Yes, but then there will be a bunch of issues with the CentOS 7 release, or even the Ubuntu 14.04 LTS release which would require a bunch of work ongoing and a MASSIVE effort to begin with to get it to the point you can even start to look at stabilising. It too and will be ongoing and you get to the end of it and you've got a bunch of awesome support for an ancient release and by then you're basically screwed. Ultimately no one will put up that time either. Someone needs to pay some how whether it be in development or money. The fact is we are in this situation because OLPC the organisation has zero interest in the project now and the organisations around the globe want everything provided to them on a plate with everything the want provided for free with the click of a finger. The fact if they all provided a little bit of development resources to the upstream project there wouldn't be this problem. Many hands and all that... > That is my concern. If we would had one dsd involved, > the conversation would be completely different, > But as Samuel said in a previous mail in this thread "I have seen a fair > amount of interest, > both publicly and privately, for newer XO laptop builds. But I don't think > the requesters > realize how much work it takes to make one." Exactly! Nor do they want to pay for the effort. There's is a HUGE amount of initial work to get everything moved to a LTS platform. CentOS would be easier in that in a lot of cases it would be recompiling packages. But there would also need to be kernel and other work which ever route you go. It would be months of work to get a distro working then you need to QA etc. Who is going to do the work, who is going to pay. No one will come out publicly and say it but they all want a polished LTS release without having to contribute any resources themselves. "Please may I have a rainbow pooing unicorn" In Fedora there is work but it's small work packages on going. I'll send a different email outlining the work I think there is there. >> > Let me explain that last point. >> > >> > There is a continuous flow of changes into Fedora. These changes >> > eventually flow into Red Hat Enterprise Linux, and thus into CentOS. >> > >> > The most cost effective way to handle this flow was for developers to >> > test changes on our builds, every week. This gaves us awareness of >> > the change and kept us involved to resist changes that cause damage. >> > We were there once. It required a low but continuous engineering >> > effort. >> >> It use to take around an hour to cut a release from Fedora/Sugar >> repos. Quite often the delta from a patch for a fix being created and >> a new OS was in the hours timeframe. It's the usual story of a little >> bit of effort regularly stops it from being a major i
Re: [Sugar-devel] Community XO software builds
Hi Peter, On Thu, May 7, 2015 at 6:16 AM, Peter Robinson wrote: > On Thu, May 7, 2015 at 12:10 AM, James Cameron wrote: > > On Wed, May 06, 2015 at 09:19:45AM -0300, Gonzalo Odiard wrote: > >> I think we should try make a build using CentOS. I don't know if > >> have all the packages we need, but the rate of change in Fedora was > >> difficult to follow when OLPC had a team dedicated and now is almost > >> impossible. The true is we didn't finished to solve the problems we > >> found in F20, and Fedora is working in F22. > > > Let make my comment clear. My proposal was not a criticize Fedora or the Fedora community. Fedora has been very supportive and responsive. > > I do not think we should switch from Fedora to CentOS, because; > > > > 1. our installed base express interest in Fedora or Ubuntu, > > Daniel Drake, myself and others put in a lot of effort back in the > F-14/15 days to get everything upstream into Fedora. I continue to > maintain that and produce a Sugar on a Stick release with every Fedora > release. > > In the last release Daniel and I was involved in the delta between > Fedora and the OLPC release was very minimal. Basically kernel, > firmware, and some minor changes to a couple of Sugar packages for XO > HW and patches that weren't yet upstream. > > > 2. there are missing desktop packages, which means we are taking on > > maintenance of those packages on CentOS, > > Having tried and failed to do this back when EL6 was new I believe > this is a dead end. It turned out to be _WAY_ more effort than > actually keeping Fedora up to date. The upstream RHEL releases are > every 6 months but if you need a fix for a package in the core 2500 > odd packages and it's not easy you might be waiting a lot longer for a > fix. > > Ok. I didn't know that. When I talked with deployments and they ask for Ubuntu, and I ask why, what they really want is Long Time Support. No deployment change their image more than once a year. In fact, change a image is a logistic challenge for most of the big/middle size deployments. Then, I was thinking in CentOS as a LTS version of Fedora. > In Fedora if you know the right people (like me) you can get a fix > into update-testing in a day. Also there's a much much wider QA group > across the packages we use and care about. > > I can go on and on about the details required for this but basically I > suspect eyes have glazed over already. > > This is true, and I know that. But also is true, that keep the pace of changes in Fedora is not easy. In fact, is not Fedora fault, mostly is Gtk ([1], [2], [3]) or libraries (the last was vte [4], but I can find more). [1] https://github.com/sugarlabs/sugar-artwork/commit/27fac30cb028a7461f40da6765db13c017ad6f13 [2] https://github.com/sugarlabs/sugar-artwork/commit/f87d4b05a2b2db55dc4a8dddc9321ac8fbe33f3e [3] https://github.com/sugarlabs/sugar-artwork/commit/e6f3c4430477176750b4ae4a007e98837a877080 [4] https://github.com/godiard/terminal-activity/commit/074f11cc37c6fa1035e32bc4132c6371254fa0f8 > 3. we would delay necessary work until the next release of CentOS, or > > if the work is too large we may never upgrade. > > I suspect it would be never. > Ok. But Let me explain my reasons. Right now, the only "stable" images are based on F18. We don't have images in a good shape for the deployments for F20, we missed F21 (where Gtk theme change in a subtle way again, and toggle toolbar buttons don't change the background color), and we should start to work in F22. With the hands we have today, I am sure we will not solve all the problems we already have before F23 is released. That is my concern. If we would had one dsd involved, the conversation would be completely different, But as Samuel said in a previous mail in this thread "I have seen a fair amount of interest, both publicly and privately, for newer XO laptop builds. But I don't think the requesters realize how much work it takes to make one." > > > Let me explain that last point. > > > > There is a continuous flow of changes into Fedora. These changes > > eventually flow into Red Hat Enterprise Linux, and thus into CentOS. > > > > The most cost effective way to handle this flow was for developers to > > test changes on our builds, every week. This gaves us awareness of > > the change and kept us involved to resist changes that cause damage. > > We were there once. It required a low but continuous engineering > > effort. > > It use to take around an hour to cut a release from Fedora/Sugar > repos. Quite often the delta from a patch for a fix being created and > a new OS was in the hours timeframe. It's the usual story of a little > bit of effort regularly stops it from being a major issue. > > Kernel and olpc-os-builder aside I think you could probably produce a > working image of Fedora 22 now. I think all the userspace bits are > likely there and working due to my SoaS work. > > I am sure we could produce a _almost_working_ image for Fedora 22. The problem i
Re: [Sugar-devel] Community XO software builds
On Wed, May 06, 2015 at 11:21:35PM -0400, Samuel Greenfeld wrote: > OLPC already appears to be going the Ubuntu LTS route When/where can I read more about what OLPC is doing with Ubuntu LTS? Apologies for the lazyweb request. Martin pgpmmvPQ1vjFZ.pgp Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 07, 2015 at 10:16:42AM +0100, Peter Robinson wrote: > On Thu, May 7, 2015 at 12:10 AM, James Cameron wrote: > > 2. there are missing desktop packages, which means we are taking on > > maintenance of those packages on CentOS, > > Having tried and failed to do this back when EL6 was new I believe > this is a dead end. It turned out to be _WAY_ more effort than > actually keeping Fedora up to date. The upstream RHEL releases are > every 6 months but if you need a fix for a package in the core 2500 > odd packages and it's not easy you might be waiting a lot longer for a > fix. > > In Fedora if you know the right people (like me) you can get a fix > into update-testing in a day. Also there's a much much wider QA group > across the packages we use and care about. I'm sure core people get it, but I think it's hard to over-emphasize to everyone else that there are two places where you get the most bang for your buck: 1) you stay with the latest (Fedora); or 2) you *never* change anything, ever. Everything in-between seems like it might be a better tradeoff, but really all that's happening is you're giving your paid devops staff time to work around their holidays and internally-driven priorities. Have no paid devops staff or worldwide priority list? You need to be on Fedora or *never* ever change. SugarLabs being in that place, with people like you to take forward Sugar packages on the popular, RHEL-upstream (in practice) Fedora, there is no good reason to accept a slower security fix process and a much more time-expensive release process. > And I've been trying as hard with Fedora as possible. The core Sugar > stack is in pretty good shape. There's some work needed on some > Activies but most of the work it to update them to the latest upstream > bits. This rings true to me too. > Peter Martin pgpETbOCXKw8f.pgp Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 7, 2015 at 4:21 AM, Samuel Greenfeld wrote: > The obvious counterargument would be that a deployment might want to deploy > your XO-Next (whatever it is) alongside existing XO laptops, allowing all of > them to have the same configuration. From my memory of olpc-os-builder it was very modular and wouldn't be hard to add dozens of different devices support to it. > There's plenty of blame to go around in terms of re-inventing the wheel and > lack of communication. Yep! > There simply (and correct me if I'm wrong) are not the resources inside of > OLPC, outside, or combined at this time to maintain and update two separate > builds & build systems. > > It amazes me how far we bend over backwards to avoid saying "end of life" > and "end of support". > > > I have seen a fair amount of interest, both publicly and privately, for > newer XO laptop builds. But I don't think the requesters realize how much > work it takes to make one. The big one here is kernel kernel kernel. > And I do not forsee anyone stepping up to get the XO-1.75 and XO-4 kernel & > drivers into a state they can be upstreamed or upgraded for newer Fedoras > unless a deployment really wants this instead of newer equipment. Or even the 1.5, I believe most of the XO-1 support is upsteream. > Newer operating systems tend to require more disk space and RAM than the > predecessors. We have seen this even within Fedora's lineage. Yes, and no. I mean 1Gb of the original XO-1 is tight, but SoaS still happily fits in 4Gb with a bunch of space to spare. Looking at my current SoaS VM the used space is around 1.9Gb. Amusingly the various cloud/container enterprise initiatives actively help us here because for once they care about dependency bloat too :-) The two things that add bloat to the current SoaS image are: * Browse needs to be converted to the new WebKitGtk APIs so we don't ship two copies of WebKitGtk. * Conversion of remaining gstreamer 0.10 to 1.0 to allow us not to ship that. Ultimately I think you could with a little development effort get it down to 1.5Gb used space which would make a 2Gb filesystem quite usable. > Since OLPC already appears to be going the Ubuntu LTS route, I would argue > it would be easiest to take everything that way, porting utilities as > required, and make that the final image & build system for XOs. Personally I have no interest in that. I wish you luck. > I only have a limited number of hours per week I can look into OLPC things, > but I'm tempted to take a look. > > > > > > > On Wed, May 6, 2015 at 10:50 PM, James Cameron wrote: >> >> On Wed, May 06, 2015 at 09:29:46PM -0400, Samuel Greenfeld wrote: >> > It might be possible for this new builder to be eventually taught to >> > handle XOs. >> >> There was no significant interest in my previous builder uxo, which >> already knows how to handle XOs. The recent posts on devel@ of people >> trying something similar without looking at uxo is further evidence of >> that. >> >> So for the moment, there seems to be no need for my new builder to >> handle XO-1, XO-1.5, XO-1.75 or XO-4 laptops. The Fedora based >> builder is working fine for those laptops. >> >> -- >> James Cameron >> http://quozl.linux.org.au/ > > > > ___ > Devel mailing list > Devel@lists.laptop.org > http://lists.laptop.org/listinfo/devel > ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Thu, May 7, 2015 at 12:10 AM, James Cameron wrote: > On Wed, May 06, 2015 at 09:19:45AM -0300, Gonzalo Odiard wrote: >> I think we should try make a build using CentOS. I don't know if >> have all the packages we need, but the rate of change in Fedora was >> difficult to follow when OLPC had a team dedicated and now is almost >> impossible. The true is we didn't finished to solve the problems we >> found in F20, and Fedora is working in F22. > > I do not think we should switch from Fedora to CentOS, because; > > 1. our installed base express interest in Fedora or Ubuntu, Daniel Drake, myself and others put in a lot of effort back in the F-14/15 days to get everything upstream into Fedora. I continue to maintain that and produce a Sugar on a Stick release with every Fedora release. In the last release Daniel and I was involved in the delta between Fedora and the OLPC release was very minimal. Basically kernel, firmware, and some minor changes to a couple of Sugar packages for XO HW and patches that weren't yet upstream. > 2. there are missing desktop packages, which means we are taking on > maintenance of those packages on CentOS, Having tried and failed to do this back when EL6 was new I believe this is a dead end. It turned out to be _WAY_ more effort than actually keeping Fedora up to date. The upstream RHEL releases are every 6 months but if you need a fix for a package in the core 2500 odd packages and it's not easy you might be waiting a lot longer for a fix. In Fedora if you know the right people (like me) you can get a fix into update-testing in a day. Also there's a much much wider QA group across the packages we use and care about. I can go on and on about the details required for this but basically I suspect eyes have glazed over already. > 3. we would delay necessary work until the next release of CentOS, or > if the work is too large we may never upgrade. I suspect it would be never. > Let me explain that last point. > > There is a continuous flow of changes into Fedora. These changes > eventually flow into Red Hat Enterprise Linux, and thus into CentOS. > > The most cost effective way to handle this flow was for developers to > test changes on our builds, every week. This gaves us awareness of > the change and kept us involved to resist changes that cause damage. > We were there once. It required a low but continuous engineering > effort. It use to take around an hour to cut a release from Fedora/Sugar repos. Quite often the delta from a patch for a fix being created and a new OS was in the hours timeframe. It's the usual story of a little bit of effort regularly stops it from being a major issue. Kernel and olpc-os-builder aside I think you could probably produce a working image of Fedora 22 now. I think all the userspace bits are likely there and working due to my SoaS work. It's actually the thing that annoys me most about the sugar community. IMO we have a great working Sugar release that works pretty much everywhere plus is a great proven base for XO releases yet so many core developers have told me "if only you'd focus on Ubuntu we'd use it" yet Ubuntu for _YEARS_ have shown that they couldn't given a shit and even actively remove core bits needed (remember the Browse on Mozilla years anyone??) to make it even harder. > The next most cost effective way is to do this work only when a new > release of Fedora occurs. This results in lots of head scratching and > bug fixing, and new builds, until the bugs are mostly gone. We are > here now. It requires bursts of engineering effort. Actually it needs work _BEFORE_ a new release happens, any work now IMO should be focused on Fedora 23. That way you have everything in place in time for Fedora 23 GA in October and you get the longest value out of the release. > The least cost effective way is to hold off doing that work for three > years until the next CentOS release. This would be a lot more work in > a much shorter burst. And you'll likely end up in a very disparate stability across devices. Both ARMv7 and i686 is community supported in CentOS which means you get likely dubious quality of work and I suspect due to toolchain config choices for i686 it won't even run on the XO-1. Has anyone actually tried booting CentOS-7 on a XO1? From what I've seen of the ARMv7 efforts I see it as half arsed at best. People ask me if I can help with CentOS. The answer is no. I have no personal interest in CentOS. I have enough to do with personal projects on Fedora. > Delaying effort until a future time hasn't worked, and I don't think > it will. Meanwhile, I'm trying as hard as I can with what I'm doing. And I've been trying as hard with Fedora as possible. The core Sugar stack is in pretty good shape. There's some work needed on some Activies but most of the work it to update them to the latest upstream bits. Peter https://alt.fedoraproject.org/pub/alt/stage/22_TC2/Images/armhfp/Fedora-SoaS-armhfp-22-TC2-sda.raw.xz https:
Re: [Sugar-devel] Community XO software builds
On Thu, May 07, 2015 at 12:56:32AM -0500, Sebastian Silva wrote: > > On 06/05/15 18:10, James Cameron wrote: > > 1. our installed base express interest in Fedora or Ubuntu, > > I wonder how accurate this is? No idea. > Many times "upstream", or suppliers, tend to lump entire deployments > into one person or group. > Did you mean administrators, technicians, teachers, or users? Yes. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On 06/05/15 18:10, James Cameron wrote: > 1. our installed base express interest in Fedora or Ubuntu, I wonder how accurate this is? Many times "upstream", or suppliers, tend to lump entire deployments into one person or group. Did you mean administrators, technicians, teachers, or users? -- I+D SomosAzucar.Org "icarito" #somosazucar en Freenode IRC "Nadie libera a nadie, nadie se libera solo. Los seres humanos se liberan en comunión" - P. Freire ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Wed, May 06, 2015 at 11:21:35PM -0400, Samuel Greenfeld wrote: > The obvious counterargument would be that a deployment might want to > deploy your XO-Next (whatever it is) alongside existing XO laptops, > allowing all of them to have the same configuration. I don't think that's likely. And if it is required, the same set of Sugar activities or the same user level desktop will suffice. The software layers are well isolated; there's no reason to have the same things all the way down to the kernel. > Since OLPC already appears to be going the Ubuntu LTS route, I would > argue it would be easiest to take everything that way, porting > utilities as required, and make that the final image & build system > for XOs. We'll see. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
The obvious counterargument would be that a deployment might want to deploy your XO-Next (whatever it is) alongside existing XO laptops, allowing all of them to have the same configuration. There's plenty of blame to go around in terms of re-inventing the wheel and lack of communication. There simply (and correct me if I'm wrong) are not the resources inside of OLPC, outside, or combined at this time to maintain and update two separate builds & build systems. It amazes me how far we bend over backwards to avoid saying "end of life" and "end of support". I have seen a fair amount of interest, both publicly and privately, for newer XO laptop builds. But I don't think the requesters realize how much work it takes to make one. And I do not forsee anyone stepping up to get the XO-1.75 and XO-4 kernel & drivers into a state they can be upstreamed or upgraded for newer Fedoras unless a deployment really wants this instead of newer equipment. Newer operating systems tend to require more disk space and RAM than the predecessors. We have seen this even within Fedora's lineage. Since OLPC already appears to be going the Ubuntu LTS route, I would argue it would be easiest to take everything that way, porting utilities as required, and make that the final image & build system for XOs. I only have a limited number of hours per week I can look into OLPC things, but I'm tempted to take a look. On Wed, May 6, 2015 at 10:50 PM, James Cameron wrote: > On Wed, May 06, 2015 at 09:29:46PM -0400, Samuel Greenfeld wrote: > > It might be possible for this new builder to be eventually taught to > > handle XOs. > > There was no significant interest in my previous builder uxo, which > already knows how to handle XOs. The recent posts on devel@ of people > trying something similar without looking at uxo is further evidence of > that. > > So for the moment, there seems to be no need for my new builder to > handle XO-1, XO-1.5, XO-1.75 or XO-4 laptops. The Fedora based > builder is working fine for those laptops. > > -- > James Cameron > http://quozl.linux.org.au/ > ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Wed, May 06, 2015 at 09:29:46PM -0400, Samuel Greenfeld wrote: > It might be possible for this new builder to be eventually taught to > handle XOs. There was no significant interest in my previous builder uxo, which already knows how to handle XOs. The recent posts on devel@ of people trying something similar without looking at uxo is further evidence of that. So for the moment, there seems to be no need for my new builder to handle XO-1, XO-1.5, XO-1.75 or XO-4 laptops. The Fedora based builder is working fine for those laptops. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
I will try to answer some questions. But my last two points will only raise new ones. 1. There are a few purposes for the community build. The first is that for a while, all the OLPC builds announced seemed to be private ones available upon request. It therefore was necessary to see if builds were still possible without the private extensions, and how well they worked. A second is that I actually build against a local mirror. This mirror was created back when it was uncertain if OLPC would keep the MIT servers, and expanded when said servers started running into problems. (These issues have since been resolved.) OLPC has and may still have an automatic backup; but I recall others having to stop it from accidentally pulling corrupted data in the past. Although it only has a subset of the public OLPC content available anonymously, building against my mirror makes sure that it still works and that it is periodically made up to date. 2. I have uploaded the .ini files I use to http://www.greenfeld.org/xo/community/builds/14.1.0/olpc-os-builder/ . But there is nothing in them that you could not derive from the olpc-14.1.0*.ini files already in olpc-os-builder. The .zd SD card image for XO-1 build 2 is next to it's .img file. I added a pause at the end of kspost.75.install_bundles.inc so I can tweak the XO-1 image and remove some of the larger activities. But this is temporary for debugging only. 3. Since I am based in the US, I cannot generate images with the multimedia items due to patents. At best I could give you instructions similar to how OLPC already does. I vaguely recall all XO-4's might be licensed for many multimedia codecs but it would be up to OLPC to make those images more widely available. 4. Personally I would argue that a CentOS or another long-term build may be the best approach for XOs. Sugar is in EPEL 6, and likely could be added to EPEL 7. It should surprise no one at this point that the list of personnel on OLPC's web site is years out of date. There may be more people working on XSCE at the moment than XO laptop software. Given the lack of personnel and resources I believe it would be best to do one final build for XO-1 through XO-4 based on a LTS distribution supported to at least 2020, and then only minor security/fixes after that. 5. OLPC already is looking beyond the XO, and beyond Fedora. If you look at dev.laptop.org closely, you might notice a bunch of tickets targeted for "su-15.1" as well as a new olpc-ubuntu-sugar-builder git tree meant for standard PCs. This appears to be an Sugar 0.104/Ubuntu 14.04 LTS build with anti-theft provided by a secure-boot-based EFI bootloader, not Open Firmware. While I am not thrilled that this has been done without the historical community's involvement, it likely matches the need of the XO Infinity or another client who currently pays the bills. It might be possible for this new builder to be eventually taught to handle XOs. But if OLPC is looking beyond the XO-4, perhaps it's time that Sugar do so as well. More information can be found at http://dev.laptop.org/ticket/12881 On Wed, May 6, 2015 at 7:18 PM, James Cameron wrote: > On Wed, May 06, 2015 at 10:49:47AM -0400, Adam Holt wrote: > > They all seem to want a better browser and better codec support to > > view various+sundry videos, within Sugar ideally, but if that's not > > possible then within Gnome. One group per week asks me for the > > above, above all else (often more than one deployment/group per > > week). > > Why isn't this reaching me and the people who would do something about > it? Please count these requests, deidentify and aggregate them, and > report them monthly on devel@ or sugar-devel@ > > > But if CentOS is not realistically achievable, F22 might be more > > appropriate, given it's final freeze is supposed to be less than 1 > > week away? > > The size of this task (F22) has not yet been estimated, but based on > Samuel's write up, my guess is between 10 and 50 engineer hours. > > There may be other problems lurking. > > The Fedora 20 port just on XO-4 has consumed way more than this. > > -- > James Cameron > http://quozl.linux.org.au/ > ___ > Sugar-devel mailing list > sugar-de...@lists.sugarlabs.org > http://lists.sugarlabs.org/listinfo/sugar-devel > ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Wed, May 06, 2015 at 10:49:47AM -0400, Adam Holt wrote: > They all seem to want a better browser and better codec support to > view various+sundry videos, within Sugar ideally, but if that's not > possible then within Gnome. One group per week asks me for the > above, above all else (often more than one deployment/group per > week). Why isn't this reaching me and the people who would do something about it? Please count these requests, deidentify and aggregate them, and report them monthly on devel@ or sugar-devel@ > But if CentOS is not realistically achievable, F22 might be more > appropriate, given it's final freeze is supposed to be less than 1 > week away? The size of this task (F22) has not yet been estimated, but based on Samuel's write up, my guess is between 10 and 50 engineer hours. There may be other problems lurking. The Fedora 20 port just on XO-4 has consumed way more than this. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
On Wed, May 06, 2015 at 09:19:45AM -0300, Gonzalo Odiard wrote: > I think we should try make a build using CentOS. I don't know if > have all the packages we need, but the rate of change in Fedora was > difficult to follow when OLPC had a team dedicated and now is almost > impossible. The true is we didn't finished to solve the problems we > found in F20, and Fedora is working in F22. I do not think we should switch from Fedora to CentOS, because; 1. our installed base express interest in Fedora or Ubuntu, 2. there are missing desktop packages, which means we are taking on maintenance of those packages on CentOS, 3. we would delay necessary work until the next release of CentOS, or if the work is too large we may never upgrade. Let me explain that last point. There is a continuous flow of changes into Fedora. These changes eventually flow into Red Hat Enterprise Linux, and thus into CentOS. The most cost effective way to handle this flow was for developers to test changes on our builds, every week. This gaves us awareness of the change and kept us involved to resist changes that cause damage. We were there once. It required a low but continuous engineering effort. The next most cost effective way is to do this work only when a new release of Fedora occurs. This results in lots of head scratching and bug fixing, and new builds, until the bugs are mostly gone. We are here now. It requires bursts of engineering effort. The least cost effective way is to hold off doing that work for three years until the next CentOS release. This would be a lot more work in a much shorter burst. Delaying effort until a future time hasn't worked, and I don't think it will. Meanwhile, I'm trying as hard as I can with what I'm doing. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
Hi Sam, > >- Who actually is using/testing these images? > > I downloaded it (XO-1 and XO-4 versions). > >- Why? > > To test if all is working in a new Fedora, and to try find a solution for the Browse problems in the XO-1. Sadly wifi connectivity is not working ok in the F20 images. > >- Is there a reason you are not looking into using an official (OLPC >or deployment) build? > > For distribution, today is more stable the F18 version. But we need move then we need solve the problems we find in newer versions. > >- Have you engaged OLPC or another party to work on changes? > > Yes. > >- What direction do you believe the builds should go? > > I think we should try make a build using CentOS. I don't know if have all the packages we need, but the rate of change in Fedora was difficult to follow when OLPC had a team dedicated and now is almost impossible. The true is we didn't finished to solve the problems we found in F20, and Fedora is working in F22. Building XO builds by repacking existing work is relatively trivial. > > But the low-level kernel, driver, and OS work necessary to support XOs > with newer operating systems (as well as newer XO batteries) is something I > cannot do, and where we really need help. > +1 > Without guidance from OLPC or others, I could build thousands of XO-# > laptop images. But unless it looks like a significant number of > deployments/children actually would benefit, there really is no point. > > I think the benefit is provide a environment where we can test, fill bugs, etc. But if there are no people with the knowledge and the time to work in the low level stuff, will be difficult. Gonzalo ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Community XO software builds
Hi Samuel, I think your volunteer work is important. It is not clear to me exactly what the focus is of your images, nor where the repositories with the ini files for the builder, or the download link for the ready images. This is probably the reason you have so few downloads logged. Perhaps we should put all of this in a Wiki page? I am interested, but haven't been able to test because when I asked last time for the SD card images you told me you would build them but never let me know when/where I could get them. I understand it may be frustrating to work without feedback but it's simply the way it works at this point, unless you are in the field. Also, keep in mind that deployments are interested in updating OS images once every one or two years. Thanks, in the name of the children, for the work you do. I'll try to respond your questions inline below. Regards, Sebastian On 05/05/15 23:54, Samuel Greenfeld wrote: > I saw some discussion last week about the community XO software builds. > > This seems to be something which gets many people excited. > > However according to my web server, there have not been very many > downloads of them. > > If I may ask: > > * Who actually is using/testing these images? > Not me, yet. > > * Why? > I maintained in the past official images for Peru. > > * Is there a reason you are not looking into using an official (OLPC > or deployment) build? > Yes we like to roll our own to include native languages, features (e.g. Sugar Network), etc. > > * Have you engaged OLPC or another party to work on changes? > I try to work upstream. > > * What direction do you believe the builds should go? > The best possible experience for end users. Basically, on XO, means performance tuning. > > Building XO builds by repacking existing work is relatively trivial. > > But the low-level kernel, driver, and OS work necessary to support XOs > with newer operating systems (as well as newer XO batteries) is > something I cannot do, and where we really need help. > > Without guidance from OLPC or others, I could build thousands of XO-# > laptop images. But unless it looks like a significant number of > deployments/children actually would benefit, there really is no point. > > --- > SJG > > > > ___ > Sugar-devel mailing list > sugar-de...@lists.sugarlabs.org > http://lists.sugarlabs.org/listinfo/sugar-devel -- I+D SomosAzucar.Org "icarito" #somosazucar en Freenode IRC "Nadie libera a nadie, nadie se libera solo. Los seres humanos se liberan en comunión" - P. Freire ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel