Status of ASF Ubuntu Buildbots
Hi All, I’ve been working on creating replacement VMs for the deprecated Tethys (10.04 64bit) , bb-vm2 (12.04 32bit) and bb-vm3 (12.04 32bit). This work is now complete. 3 VMs replaced with 2 x 14.04 LTS VMs. (As an aside there is a 16.04 LTS buildbot available if you want to do tests on it) 3 builds were moved from Tethys to bb_slave4_ubuntu - the linux64 nightly build and the two RAT builds (trunk and aoo410 testing) All 3 builds are successful currently. The linux32-nightly and linux32-snapshot builds have been moved from bb-vm2 and bb-vm3 to bb_slave5_ubuntu - initial builds are still being performed. I reduced the frequency of some builds - I mean why spend 9 hours building and uploading language packs every single day, there is no need - especially from the aoo410 branch thats seen like 2 commits in 9 months! the openoffice.conf file has been updated. Feel free to start tweaking the linux32 builds after the initial builds have completed in a day or so. Any questions, I’m here on list. Gav… - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Question about binary upload process
Patricia Shanahan wrote: I suspect if I try to personally push all the artifacts, it will take a lot longer because of the limitations of my home PCs, and cable supplier. I'm supposed to get "up to" 150 Mbps, but I don't think I get that all the time. If your upload speed is in that order of magnitude, then you are more than OK; at ~100 Mbps (if this is your upload rate) you would be able to upload 750 MBytes per minute, absolutely OK for the size of our artifacts. The big unknown remains the ASF server speed. That said, you needn't upload all artifacts yourself. That is one option. But with just a bit of organization we can split the upload between multiple PMC members (only PMC members can commit to that directory). Regards, Andrea. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Question about binary upload process
On 9/9/2016 3:23 PM, Andrea Pescetti wrote: Marcus wrote: Am 09/09/2016 04:38 PM, schrieb Dave Brondsema: On 9/8/16 6:31 PM, Andrea Pescetti wrote: 2) Whoever is in the best position to do so, uploads the binaries to the ASF. This can be done also by multiple people, who upload to different subdirs here: https://dist.apache.org/repos/dist/dev/openoffice/ I'll expand a bit more: this is done through SVN commit. This page http://www.apache.org/dev/release-publishing has up-to-date information, and it's likely Patricia already stumbled upon it. If you need a more precise estimate, the upload of artifacts took about 48 hours; the upload of the same artifacts to SourceForge took about 2 hours. I hope (and kind of assume) that the ASF servers performance is now better. I suspect if I try to personally push all the artifacts, it will take a lot longer because of the limitations of my home PCs, and cable supplier. I'm supposed to get "up to" 150 Mbps, but I don't think I get that all the time. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Code signing available for OpenOffice
Mark Thomas wrote: The infrastructure team has regained access to the OpenOffice code signing account. If you would like to use it to sign releases please open an infra ticket and provide the Apache IDs of those committers that need access. May I ask if this is the same Symantec system that was rumored to be about to be abandoned in early 2016? I hadn't renewed credentials since it seemed clear that the ASF would abandon it. Anyway, thanks for getting access again; just to be clear, it has not been discussed so far to use it very soon, but it's good to know that the ASF still has signing facilities available. Regards, Andrea. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Question about binary upload process
Marcus wrote: Am 09/09/2016 04:38 PM, schrieb Dave Brondsema: On 9/8/16 6:31 PM, Andrea Pescetti wrote: 2) Whoever is in the best position to do so, uploads the binaries to the ASF. This can be done also by multiple people, who upload to different subdirs here: https://dist.apache.org/repos/dist/dev/openoffice/ I'll expand a bit more: this is done through SVN commit. This page http://www.apache.org/dev/release-publishing has up-to-date information, and it's likely Patricia already stumbled upon it. If you need a more precise estimate, the upload of artifacts took about 48 hours; the upload of the same artifacts to SourceForge took about 2 hours. I hope (and kind of assume) that the ASF servers performance is now better. Additionally, as the files are rsync'd up to SourceForge, they start getting pushed out to SourceForge's mirrors which can take some time too. You can click the "i" icon for a file to see how many mirrors it is on so far. In the 4.1.2 case I was quite surprised in positive since the files were immediately available for download. Then I don't recall now if I activated (with instructions from Marcus) the flag for "staging" it, but it was discussed back at the time. Let me add that Infra provided good support for 4.1.2, especially for RC1 when we needed some significant configuration changes to accommodate our RC. These changes are now permanent. Let me also add that that a major breakthrough there was that I was able to sit with Infra during ApacheCon and explore multiple solutions. Seeing the little interest in attending ApacheCon now, I think we can't rely on an ApacheCon sprint now. Though, I think no changes are needed at the moment. Regards, Andrea. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Differentiate or Die
On 09/09/2016 18:11, Jim Jagielski wrote: > We should be in touch with what our users, and our potential users, want. Do you mean something other https://bz.apache.org/ooo/buglist.cgi?f1=votes=greaterthan=Bug%20Number_format=advanced=---=100 jonathon - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Differentiate or Die
Le 09/09/2016 à 20:11, Jim Jagielski a écrit : One of the great things about FOSS is the tight connection between users and developers. After all, most developers are users that have an itch to scratch. If there are things that the user community wants, then chances are good that developers will be jazzed about working on them, or, at least, the pool of potential developers might be increased. But open source, and open source projects, should not be run in a normal, corporate s/w development mode, where some "entity" decides what features are needed, etc... We should be in touch with what our users, and our potential users, want. No need to go very far. There is a bug tracker with Requests For Enhancements and votes. Hagar - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Code signing available for OpenOffice
OpenOffice developers, The infrastructure team has regained access to the OpenOffice code signing account. If you would like to use it to sign releases please open an infra ticket and provide the Apache IDs of those committers that need access. Note that you can test sign files as much as you like but production signings cost the ASF real $$$. If you have any questions, please do get in touch. Cheers, Mark - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Differentiate or Die
One of the great things about FOSS is the tight connection between users and developers. After all, most developers are users that have an itch to scratch. If there are things that the user community wants, then chances are good that developers will be jazzed about working on them, or, at least, the pool of potential developers might be increased. But open source, and open source projects, should not be run in a normal, corporate s/w development mode, where some "entity" decides what features are needed, etc... We should be in touch with what our users, and our potential users, want. > On Sep 9, 2016, at 1:53 PM, Jorg Schmidtwrote: > >> From: Jim Jagielski [mailto:j...@jagunet.com] > >>> LibreOffice has a list of big ideas, called "crazy ideas", at >>> https://wiki.documentfoundation.org/Development/Crazy_Ideas >>> These require big effort and it would be great if an office suite >>> would implement them. >>> Notable examples are >>> 1. multi process instances >>> 2. split MSOffice support in library >>> >>> Picking one of those and implementing it, would allow to >> differentiate. >>> >> >> Why not ask our user community? > > Yes, that would be a theoretically good way, but I fear that's in practice a > very > complicated subject. > > Let me formulate in short: open source communities work mostly meritocratic, > not > democratic > > > Jorg > > > > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Differentiate or Die
> On Sep 9, 2016, at 11:31 AM, Simos Xenitellis> wrote: > > > LibreOffice has a list of big ideas, called "crazy ideas", at > https://wiki.documentfoundation.org/Development/Crazy_Ideas > These require big effort and it would be great if an office suite > would implement them. > Notable examples are > 1. multi process instances > 2. split MSOffice support in library > > Picking one of those and implementing it, would allow to differentiate. > Why not ask our user community? - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Question about binary upload process
Am 09/09/2016 04:38 PM, schrieb Dave Brondsema: On 9/8/16 6:31 PM, Andrea Pescetti wrote: Patricia Shanahan wrote: we have a lot of binaries to upload. Could someone with experience or knowledge of the process tell me a bit about how it is done, how long it takes, and what, if anything, it costs ASF? Sure. This changed just days before 4.1.2, but it still holds. 1) You create the binaries. These might be created by different people too. At the end, you have a few dozen Gigabytes. Note: this must be done for every release candidate; we had 3 for 4.1.2; I recommend choosing things/issues wisely so that 4.1.3 can aim at having only 1 RC (i.e., getting the first one right). 2) Whoever is in the best position to do so, uploads the binaries to the ASF. This can be done also by multiple people, who upload to different subdirs here: https://dist.apache.org/repos/dist/dev/openoffice/ For all the 4.1.2 RCs I did it alone, from a good connection, and it took an absurd number of hours since the connection was slow on the ASF side. Speed may be better now (honestly, I don't see how speed could be worse). A good trick was to upload artifacts to people.apache.org and commit from there: this was much faster, but Infra has now disallowed it by (almost) decommissioning people.apache.org 3) Only the final one must be uploaded to SourceForge; I copy/paste from https://cwiki.apache.org/confluence/display/OOOUSERS/AOO+4.1.2 section "Upload builds to mirrors". "Volunteers: Andrea Pescetti - Copy requires just a few hours, with the normal rsync instructions shown at https://sourceforge.net/p/forge/documentation/File%20Management/ (project name is openofficeorg.mirror). Set new files as "Latest Version": done by Marcus Lange see http://s.apache.org/uaj "; you probably don't have admin permissions for the OpenOffice project on SourceForge, but all other members can give you admin access. Just ask. Additionally, as the files are rsync'd up to SourceForge, they start getting pushed out to SourceForge's mirrors which can take some time too. You can click the "i" icon for a file to see how many mirrors it is on so far. I recommend creating the 4.1.3 directory via the web interface, which will let you "stage" it, meaning it will be hidden from common visitors. After all the files are uploaded and at least several mirrors have them, you can "unstage" the directory at the official release time. If you have any issues with it, support staff via https://sourceforge.net/support should be responsive, or reach out to me directly and I'll be glad to help. oh yes, the staging feature is a great help. Thanks for your offer. We will upload the files and come back to you when we need help. Marcus 4) On dist, moving from dev to the actual tree is just a matter of svn mv. This at least is very fast. 5) Costs: we use standard ASF infrastructure here, and our own time. Costs for the ASF are just the ordinary running costs that they wpuld have even if we don't release anything. Waste of storage space due to storing in SVN hundreds of GBytes of non-approved RCs was not an issue last time I spoke to Infra about this. Let me add that Infra provided good support for 4.1.2, especially for RC1 when we needed some significant configuration changes to accommodate our RC. These changes are now permanent. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Differentiate or Die
Hi Jörg; ... > From: Pedro Giffuni [mailto:p...@apache.org] > The thing about so-called "marketing gurus" is that their assumptions > about how the markets work may break down when we are talking about > software that has zero cost. > > I will simplify the marketing issue making a bold statement: "We have > millions of users because we do 80% of what the market leader > does but > with 0% of the price." No, the success of free software is not a question of price. The success of free software is based on many things, of which price is only one of them. In the case of OpenOffice I sustain that the main factor for success is that end-users perceive it as free as in price. That and not the fact that we have less developers than other projects or that we are not being distributed in major linux distributions accounts for the project being successful today. The development model of free software is something else, but it's not free. That is not the goal. read: https://www.gnu.org/philosophy/selling.en.html The GNU copyleftists have always struggled with economics but that is rather off-topic. The notion of living on distribution costs is a dead end from a gone era. Distribution costs have diminished hugely with the Internet, in such a way that even commercial producers sell more software online than on CDs. Have you ever paid for using GCC, or do you know anyone that would prefer clang because it's cheaper to download? Nowadays, support is likely the mayor source of revenue for independent developers and publicity is the mayor source of revenue for content producers. Furthermore: The development and use of OpenOffice is not free, because developers have to be paid by their companies or donate their own time. Users have cost for installation, maintenance and staff training. Absolutely, just compiling the code involves electricity costs, but the end user doesn't have to carry the burden. Many don't even know or have to be aware of the costs involved. No one has really quantified the real cost of producing OpenOffice and then if we charged even just $1 we would more than cover what we spend in development (100 million in budget.. yay!). The work of Apache is also not free, because Apache needs donations to be able to work. For example see: https://www.apache.org/foundation/sponsorship.html on: http://www.apache.org/foundation/thanks.html you can see that the sum of the sponsorship is (currently, per year): Platinum: 700,000$ Gold: 320,000$ Silver: 260,000$ Bronze: 90,000$ Such costs existed before OpenOffice was an Apache Project, and we can't at all quantify how much OpenOffice's value is for the Apache Software Foundation. Have there been more donations thanks to OpenOffice? And then .. if Microsoft or Google make a donation to the ASF does it mean either of them is in direct support of OpenOffice? Part of the appeal of the ASF is making use of Foundation resources like buildbots and mailinglists but noways anyone can fork they own codebase on github, enable travis-ci and distribute the code through other means. Money is important everywhere but it is not an absolute truth that opensource necessarily obeys market rules. Pedro.
Re: Differentiate or Die
On Fri, Sep 9, 2016 at 5:53 PM, Phillip Rhodeswrote: > Sure, I don't claim it's a perfect analogy between "their" world and the > world of F/OSS. > But I think the broad point generalizes well enough to apply to us: > > Have *some* differentiating factor that defines why a group of people would > find your > product more desirable than the other options. > > If, for us, that is "it's almost like MS Office, but free" and we're good > with that, then that's > cool. But I kinda think we ought to stretch for something a little more > specific, especially > given that "almost like MS Office, but free" isn't a unique position. > > And I'm not proposing any big, elaborate "process" or suggesting we > radically change > directions (unless we want too!) but rather just saying we could/should > spend some > time thinking about what makes AOO special. > LibreOffice has a list of big ideas, called "crazy ideas", at https://wiki.documentfoundation.org/Development/Crazy_Ideas These require big effort and it would be great if an office suite would implement them. Notable examples are 1. multi process instances 2. split MSOffice support in library Picking one of those and implementing it, would allow to differentiate. Simos > On Fri, Sep 9, 2016 at 1:41 AM, Jörg Schmidt wrote: > >> >> > From: Pedro Giffuni [mailto:p...@apache.org] >> >> > The thing about so-called "marketing gurus" is that their assumptions >> > about how the markets work may break down when we are talking about >> > software that has zero cost. >> > >> > I will simplify the marketing issue making a bold statement: "We have >> > millions of users because we do 80% of what the market leader >> > does but >> > with 0% of the price." >> >> >> No, the success of free software is not a question of price. >> >> The development model of free software is something else, but it's not >> free. That is not the goal. >> >> read: >> https://www.gnu.org/philosophy/selling.en.html >> >> Furthermore: >> The development and use of OpenOffice is not free, because developers have >> to be paid by their companies or donate their own time. Users have cost for >> installation, maintenance and staff training. >> >> The work of Apache is also not free, because Apache needs donations to be >> able to work. >> >> For example see: >> https://www.apache.org/foundation/sponsorship.html >> >> on: >> http://www.apache.org/foundation/thanks.html >> >> you can see that the sum of the sponsorship is (currently, per year): >> >> Platinum: 700,000$ >> Gold: 320,000$ >> Silver: 260,000$ >> Bronze: 90,000$ >> >> >> >> Jörg >> >> >> - >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >> For additional commands, e-mail: dev-h...@openoffice.apache.org >> >> - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Differentiate or Die
Sure, I don't claim it's a perfect analogy between "their" world and the world of F/OSS. But I think the broad point generalizes well enough to apply to us: Have *some* differentiating factor that defines why a group of people would find your product more desirable than the other options. If, for us, that is "it's almost like MS Office, but free" and we're good with that, then that's cool. But I kinda think we ought to stretch for something a little more specific, especially given that "almost like MS Office, but free" isn't a unique position. And I'm not proposing any big, elaborate "process" or suggesting we radically change directions (unless we want too!) but rather just saying we could/should spend some time thinking about what makes AOO special. Phil This message optimized for indexing by NSA PRISM On Fri, Sep 9, 2016 at 1:41 AM, Jörg Schmidtwrote: > > > From: Pedro Giffuni [mailto:p...@apache.org] > > > The thing about so-called "marketing gurus" is that their assumptions > > about how the markets work may break down when we are talking about > > software that has zero cost. > > > > I will simplify the marketing issue making a bold statement: "We have > > millions of users because we do 80% of what the market leader > > does but > > with 0% of the price." > > > No, the success of free software is not a question of price. > > The development model of free software is something else, but it's not > free. That is not the goal. > > read: > https://www.gnu.org/philosophy/selling.en.html > > Furthermore: > The development and use of OpenOffice is not free, because developers have > to be paid by their companies or donate their own time. Users have cost for > installation, maintenance and staff training. > > The work of Apache is also not free, because Apache needs donations to be > able to work. > > For example see: > https://www.apache.org/foundation/sponsorship.html > > on: > http://www.apache.org/foundation/thanks.html > > you can see that the sum of the sponsorship is (currently, per year): > > Platinum: 700,000$ > Gold: 320,000$ > Silver: 260,000$ > Bronze: 90,000$ > > > > Jörg > > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > >
Re: Question about binary upload process
On 9/8/16 6:31 PM, Andrea Pescetti wrote: > Patricia Shanahan wrote: >> we have a lot of binaries to upload. >> Could someone with experience or knowledge of the process tell me a bit >> about how it is done, how long it takes, and what, if anything, it costs >> ASF? > > Sure. This changed just days before 4.1.2, but it still holds. > > 1) You create the binaries. These might be created by different people too. At > the end, you have a few dozen Gigabytes. Note: this must be done for every > release candidate; we had 3 for 4.1.2; I recommend choosing things/issues > wisely > so that 4.1.3 can aim at having only 1 RC (i.e., getting the first one right). > > 2) Whoever is in the best position to do so, uploads the binaries to the ASF. > This can be done also by multiple people, who upload to different subdirs > here: > https://dist.apache.org/repos/dist/dev/openoffice/ > For all the 4.1.2 RCs I did it alone, from a good connection, and it took an > absurd number of hours since the connection was slow on the ASF side. Speed > may > be better now (honestly, I don't see how speed could be worse). A good trick > was > to upload artifacts to people.apache.org and commit from there: this was much > faster, but Infra has now disallowed it by (almost) decommissioning > people.apache.org > > 3) Only the final one must be uploaded to SourceForge; I copy/paste from > https://cwiki.apache.org/confluence/display/OOOUSERS/AOO+4.1.2 section "Upload > builds to mirrors". "Volunteers: Andrea Pescetti - Copy requires just a few > hours, with the normal rsync instructions shown at > https://sourceforge.net/p/forge/documentation/File%20Management/ (project name > is openofficeorg.mirror). Set new files as "Latest Version": done by Marcus > Lange see http://s.apache.org/uaj "; you probably don't have admin permissions > for the OpenOffice project on SourceForge, but all other members can give you > admin access. Just ask. > Additionally, as the files are rsync'd up to SourceForge, they start getting pushed out to SourceForge's mirrors which can take some time too. You can click the "i" icon for a file to see how many mirrors it is on so far. I recommend creating the 4.1.3 directory via the web interface, which will let you "stage" it, meaning it will be hidden from common visitors. After all the files are uploaded and at least several mirrors have them, you can "unstage" the directory at the official release time. If you have any issues with it, support staff via https://sourceforge.net/support should be responsive, or reach out to me directly and I'll be glad to help. > 4) On dist, moving from dev to the actual tree is just a matter of svn mv. > This > at least is very fast. > > 5) Costs: we use standard ASF infrastructure here, and our own time. Costs for > the ASF are just the ordinary running costs that they wpuld have even if we > don't release anything. Waste of storage space due to storing in SVN hundreds > of > GBytes of non-approved RCs was not an issue last time I spoke to Infra about > this. > > Let me add that Infra provided good support for 4.1.2, especially for RC1 when > we needed some significant configuration changes to accommodate our RC. These > changes are now permanent. > > Regards, > Andrea. > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > -- Dave Brondsema : d...@brondsema.net http://www.brondsema.net : personal http://www.splike.com : programming <>< - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Another LWN article
Even ignoring trolls is tiring work :) - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Release process and 4.1.3
Patricia Shanahan wrote: https://cwiki.apache.org/confluence/display/OOOUSERS/Release+Planning+Template It is what I would want to do for a major release with user interface changes. Yes, that one is the template for a 4.2.0, not for a 4.1.3 release. We also need something far, far more agile for getting simple bug fix releases out quickly and easily. I propose using 4.1.3 as a test case for a stripped down process. Actually, 4.1.2 was exactly this. Your starting point should thus be https://cwiki.apache.org/confluence/display/OOOUSERS/AOO+4.1.2 ; just copy the wiki page and edit/generalize accordingly. No string changes means we do not need the "String freeze" or "Translation phase" steps. No significant changes to external interfaces, combined with a small number of relatively simple fixes, eliminates the "Beta Release" phase. This is the difference between a "micro" (third digit) and a "minor" (second digit) release. Calling it 4.1.3 implies all of this, barring exceptional situations. We still need to pick the bug fixes to go in the release, construct a release candidate, test it, write release notes etc. This is all covered in the link I sent. There might be some steps that need better clarification though. Regards, Andrea. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org