Hi Mateusz, >> I would like to update or add several packages to the SvarDOS repo at >> OSDN. But I don't want to bother you every time with grabbing packages >> from my tmp web folder, uploading, rebuilding the index, ... >> >> Would you be fine, if I could do this? > > Very much in favor, yes! The (little) problem is only technicalities > that we'd need to overcome with some method. Let me start with > explaining how it works today.
Cool! Okay. > The build process of floppy images, ISO and repository indexing is > automated through a single shell script called "build.sh", it is at the > root of the svn repo. This script does basically three things: > 1. processes packages from the directory "packages" to strip sources > 2. creates the installation floppy images (and ISO, and USB image) > 3. rebuilds the index of the web repo packages ok > The index of web repo packages is generated by the buildidx.php tool (in > the buildidx/ directory on svn). Then, once everything is > created/generated, content must be pushed to the web server using the > "sync.sh" script. ok > There is one problem, though: packages are not part of SVN, so the > "packages" directory on svn is empty. The packages are primarily on my > laptop's hard disk, and I back them up from time to time to the file > repo of OSDN: https://osdn.net/projects/svardos/storage/packages/ That's something I don't understand. Wait... Probably I understand indeed. Current process is: 1. I tell you I uploaded a package to http://www.bttr-software.de/tmp/ 2. You pull that package to your laptop to the "package" folder. 3. You run all your script stuff locally an your laptop, which will generate the web repo packages. 4. Results will be synced to OSDN. Is this correct? > Another problem is that publishing a new package leads to re-generation > of all installation disks, which is probably overkill... As we are working on a low-ressources OS, we should not require several AMD EPYC CPUs to maintain it. ;-) Installation media should be re-generated only, when CORE packages (or pcntpk/udvd2) are updated. > All this to say that the current way it works is not very > "cooperation-friendly" and I think it is a good time to evolve it. +1 > What I think could work, is following changes: > > 1. all packages are stored in SVN Um, typically one doesn't store blobs in SVN/Git. > 1b. this requires new svn hosting as OSDN does not allow non-libre stuff But with the web repo packages you already have non-libre stuff, e.g., "btitanic" at OSDN, don't know? Finding another hoster might be a bummer as long as we don't want to pay for hosting. :-( And I already feel at home at OSDN now. ;-) I mean, having some gigs of web storage somewhere plus a script running every hour to check for new uploads and rebuild whatever is needed, doesn't sound very hard to me. -- Although I never did such a thing. *g* > 2. separate web repository recomputation from the process of creating > installation images to make it as simple as possible ok > With these changes, adding or updating a package would be as simple as: > 1. svn up > 2. svn add newpackage.zip > 3. svn commit > 4. regen_web_repo.sh > 5. sync_web_repo.sh > > Then, in a future step, we could imagine that steps 4 and 5 are > performed automatically by the svn server when a new package is > commited, so the process would be even simpler. Would that be something like a Git hook action? > Does it make sense? In general, it does. :-) Cheers, Robert -- +++ BTTR Software +++ Home page: https://www.bttr-software.de/ DOS ain't dead: https://www.bttr-software.de/forum/ _______________________________________________ Svardos-users mailing list [email protected] https://lists.osdn.me/mailman/listinfo/svardos-users
