Feature changed by: Stephen Kellat (skellat) Feature #308617, revision 3 Title: provide FreeBSD, Solaris and OSX build targets Buildservice: Unconfirmed Priority Requester: Mandatory Requested by: Elmar Stellnberger (estellnb) Description: The provision of build targets for the most important Unix derivatives like FreeBSD, Solaris and OSX would greatly enhance not only the capabilities of the build service but also the value it provides. While rebuilding packages for different distributions and Linux versions is often a rather cosmetic issue this could really leverage the future use of the build service and constitute a major advancement.  This should not be impossible since there are also plans for a Windows build target.   Test Case: A package like gpodder could be built on the build service and have all of its target platforms reached instead of multiple people having to take action to serve all desired platforms now.  The packagers list for gpodder can be found at http://gpodder.org/downloads.html for review.  Tolkien referred to one ring to rule them all and extending OBS this way might start to come close. Use Case: - Open source could remove any parochialism by being able to spread - packages to the BSDs, Solaris, and MacOS X.  Some packages like Banshee - work across many flavors of Linux but presently are not supported on - the BSD-rooted MacOS X. + FLOSS could remove any appearances of parochialism by being able to + spread packages to the BSDs, Solaris, and MacOS X.  Some packages like + Banshee work across many flavors of Linux but presently are not + supported on the BSD-rooted MacOS X. -- openSUSE Feature: https://features.opensuse.org/308617

Feature changed by: Elmar Stellnberger (estellnb) Feature #308617, revision 5 Title: provide FreeBSD, Solaris and OSX build targets Buildservice: Unconfirmed Priority Requester: Mandatory Requested by: Elmar Stellnberger (estellnb) Description: The provision of build targets for the most important Unix derivatives like FreeBSD, Solaris and OSX would greatly enhance not only the capabilities of the build service but also the value it provides. While rebuilding packages for different distributions and Linux versions is often a rather cosmetic issue this could really leverage the future use of the build service and constitute a major advancement.  This should not be impossible since there are also plans for a Windows build target. Test Case: A package like gpodder could be built on the build service and have all of its target platforms reached instead of multiple people having to take action to serve all desired platforms now.  The packagers list for gpodder can be found at http://gpodder.org/downloads.html for review.  Tolkien referred to one ring to rule them all and extending OBS this way might start to come close. Use Case: FLOSS could remove any appearances of parochialism by being able to spread packages to the BSDs, Solaris, and MacOS X.  Some packages like Banshee work across many flavors of Linux but presently are not supported on the BSD-rooted MacOS X. Discussion: #1: Adrian Schröter (adriansuse) (2010-01-22 15:03:55) Yes, doable, but currently no resources available here. Speak up if someone wants to do this and needs support. + #2: Elmar Stellnberger (estellnb) (2010-03-21 16:10:28) (reply to #1) + What will it take to implement a FreeBSD or Darwin Build Target? Am I + supposed to cross-compile FreeBSD from Linux or will it suffice to + upload the binary image of a base system and to accomodate the OBS so + that it can cope with system images from other operating systems? + Additionally it may be a lot of work to make the required set of basic + libraries and packages like Xorg or perhaps also KDE compile with the + new build target. -- openSUSE Feature: https://features.opensuse.org/308617

Feature changed by: Barry Allard (nv8ra3s) Feature #308617, revision 8 Title: provide FreeBSD, Solaris and OSX build targets Buildservice: Unconfirmed Priority Requester: Mandatory Requested by: Elmar Stellnberger (estellnb) Partner organization: openSUSE.org Description: The provision of build targets for the most important Unix derivatives like FreeBSD, Solaris and OSX would greatly enhance not only the capabilities of the build service but also the value it provides. While rebuilding packages for different distributions and Linux versions is often a rather cosmetic issue this could really leverage the future use of the build service and constitute a major advancement.  This should not be impossible since there are also plans for a Windows build target. Test Case: A package like gpodder could be built on the build service and have all of its target platforms reached instead of multiple people having to - take action to serve all desired platforms now.  The packagers list for - gpodder can be found at http://gpodder.org/downloads.html for - review.  Tolkien referred to one ring to rule them all and extending - OBS this way might start to come close. + take action to serve all desired platforms now. The packagers list for + gpodder can be found at http://gpodder.org/downloads.html for review. + Tolkien referred to one ring to rule them all and extending OBS this + way might start to come close. + A first internal testcase could be as simple as 'main(){}', but + partnering+feedback with a significant high-visibility project already + on OBS onto one additional platform seems a shortest path to awesome. Use Case: FLOSS could remove any appearances of parochialism by being able to spread packages to the BSDs, Solaris, and MacOS X.  Some packages like Banshee work across many flavors of Linux but presently are not supported on the BSD-rooted MacOS X. Discussion: #1: Adrian Schröter (adriansuse) (2010-01-22 15:03:55) Yes, doable, but currently no resources available here. Speak up if someone wants to do this and needs support. #2: Elmar Stellnberger (estellnb) (2010-03-21 16:10:28) (reply to #1) What will it take to implement a FreeBSD or Darwin Build Target? Am I supposed to cross-compile FreeBSD from Linux or will it suffice to upload the binary image of a base system and to accomodate the OBS so that it can cope with system images from other operating systems? Additionally it may be a lot of work to make the required set of basic libraries and packages like Xorg or perhaps also KDE compile with the new build target. #3: Barry Allard (nv8ra3s) (2012-10-09 21:34:41) (reply to #2) This an old but timeless thread important to OBS' continued increasing awesome. Perhaps consideration of the the total impact vs. benefit to add a popular non-Linux platform, say SmartOS or FreeBSD. SmartOS supports KVM + zones + ZFS. FreeBSD has jails + ZFS, AWS (ancient+custom domU) but not Xen dom0 HVM solidly (http://wiki.freebsd.org/FreeBSD/Xen) . Either would be able to spawn clean, CoW workers with obvious benefits to budget-conscious sponsors. And, there are known interests in one of these spaces that play well with others. Heck a couple of them could be convinced to dontaining gear if needed. I'm sure others will also see the value in getting involved because it will decrease support costs for one-of, non-portable, homegrown CI white-elephants and make it many times faster to get releases ready to ship; especially up-and-coming shops that want to get moving. Also, it is only common sense that it is both simpler and easier to accept the reality of existing tools and platforms that are already quite good at what they do, rather than trying to adapt an unproven tool as otherwise suggested into trying to doing everything and nothing well, which seems to go straight against unix principles. If it's any consolation, this is how we are able to ship on Darwin, many Linices, several BSD and yes, even that other platform. Bonus: Might mention to potential patrons: "Showing-off new releases that grab headlines with an awesome new (insert obvious hint here) because customers are usually busy on their own release schedule to have time/materials to want to build something out themselves. Also, it's easy for anyone to get stuck in their own seemingly infinitely large femtocosm until someone (i.e., us) come along with awesome. Especially in a way that's just more inherently scalable. It's a no- brainer because it takes the absolute drudgery of release management out in just the same way Github did for sharing code. There's money in that. Perhaps you'd like to consider some skin in this game?" [DFTWL PHB-ease with a free pitch sans slide deck] -- openSUSE Feature: https://features.opensuse.org/308617

Feature changed by: Andreas Jaeger (a_jaeger) Feature #308617, revision 10 Title: provide FreeBSD, Solaris and OSX build targets - Buildservice: Unconfirmed + Buildservice: New Priority Requester: Mandatory Requested by: Elmar Stellnberger (estellnb) Partner organization: openSUSE.org Description: The provision of build targets for the most important Unix derivatives like FreeBSD, Solaris and OSX would greatly enhance not only the capabilities of the build service but also the value it provides. While rebuilding packages for different distributions and Linux versions is often a rather cosmetic issue this could really leverage the future use - of the build service and constitute a major advancement.  This should + of the build service and constitute a major advancement. This should not be impossible since there are also plans for a Windows build target. Test Case: A package like gpodder could be built on the build service and have all of its target platforms reached instead of multiple people having to take action to serve all desired platforms now. The packagers list for gpodder can be found at http://gpodder.org/downloads.html for review. Tolkien referred to one ring to rule them all and extending OBS this way might start to come close. A first internal testcase could be as simple as 'main(){}', but partnering+feedback with a significant high-visibility project already on OBS onto one additional platform seems a shortest path to awesome hackernews posts. Use Case: FLOSS could remove any appearances of parochialism by being able to - spread packages to the BSDs, Solaris, and MacOS X.  Some packages like + spread packages to the BSDs, Solaris, and MacOS X. Some packages like Banshee work across many flavors of Linux but presently are not supported on the BSD-rooted MacOS X. Discussion: #1: Adrian Schröter (adriansuse) (2010-01-22 15:03:55) Yes, doable, but currently no resources available here. Speak up if someone wants to do this and needs support. #2: Elmar Stellnberger (estellnb) (2010-03-21 16:10:28) (reply to #1) What will it take to implement a FreeBSD or Darwin Build Target? Am I supposed to cross-compile FreeBSD from Linux or will it suffice to upload the binary image of a base system and to accomodate the OBS so that it can cope with system images from other operating systems? Additionally it may be a lot of work to make the required set of basic libraries and packages like Xorg or perhaps also KDE compile with the new build target. #3: Barry Allard (nv8ra3s) (2012-10-09 21:34:41) (reply to #2) This an old but timeless thread important to OBS' continued increasing awesome. Perhaps consideration of the the total impact vs. benefit to add a popular non-Linux platform, say SmartOS or FreeBSD. SmartOS supports KVM + zones + ZFS. FreeBSD has jails + ZFS, AWS (ancient+custom domU) but not Xen dom0 HVM solidly (http://wiki.freebsd.org/FreeBSD/Xen) . Either would be able to spawn clean, CoW workers with obvious benefits to budget-conscious sponsors. And, there are known interests in one of these spaces that play well with others. Heck a couple of them could be convinced to dontaining gear if needed. I'm sure others will also see the value in getting involved because it will decrease support costs for one-of, non-portable, homegrown CI white-elephants and make it many times faster to get releases ready to ship; especially up-and-coming shops that want to get moving. Also, it is only common sense that it is both simpler and easier to accept the reality of existing tools and platforms that are already quite good at what they do, rather than trying to adapt an unproven tool as otherwise suggested into trying to doing everything and nothing well, which seems to go straight against unix principles. If it's any consolation, this is how we are able to ship on Darwin, many Linices, several BSD and yes, even that other platform. Bonus: Might mention to potential patrons: "Showing-off new releases that grab headlines with an awesome new (insert obvious hint here) because customers are usually busy on their own release schedule to have time/materials to want to build something out themselves. Also, it's easy for anyone to get stuck in their own seemingly infinitely large femtocosm until someone (i.e., us) come along with awesome. Especially in a way that's just more inherently scalable. It's a no- brainer because it takes the absolute drudgery of release management out in just the same way Github did for sharing code. There's money in that. Perhaps you'd like to consider some skin in this game?" [DFTWL PHB-ease with a free pitch sans slide deck] -- openSUSE Feature: https://features.opensuse.org/308617