some changes to repository and osdn to sourceforge

to avoid confusion of previous mirrorlist setup
This commit is contained in:
joborun linux 2023-07-29 10:52:19 +00:00
parent d9b74d1d3e
commit 15c41c1639

View file

@ -1,7 +1,7 @@
### [[Back to the top joborun wiki page|index.md]] ###
EDIT: February 2023 - In the gamma phase of development jobcomm has been carrying all listed packages as binaries just like jobcore & jobextra. The idea though still remains, the intention is for you to engage in building your own packages instead of the lazy way of downloading and installing binaries. In the case of browsers, other than our beloved jobextra/librewolf it takes nearly as much time to download our binary (ie palemoon - about to be dropped because enough is enough) than to download and make the package from a 3rd party's binary (palemoon...deb). The source/PKGBUILDs may remain in the jobcomm.git but may be removed from the pacman.db list.
EDIT: July 2023 - In the current phase of development jobcomm has been carrying all listed packages as binaries just like jobcore & jobextra. The idea though still remains, the intention is for you to engage in building your own packages instead of the lazy way of downloading and installing binaries. In the case of browsers, other than our beloved jobextra/librewolf it takes nearly as much time to download our binary (ie palemoon - about to be dropped because enough is enough) than to download and make the package from a 3rd party's binary (palemoon...deb). The source/PKGBUILDs may remain in the jobcomm.git but may be removed from the pacman.db list.
So the following has been altered in practice but the underlying concern remains. We have also shifted to match Arch's pkg list better and jobextra pkgs tend to match some in extra, while jobcomm to community. The ones you may otherwise find in AUR and their dependencies will most likely stay in jobcomm for the deep future and be maintained as well as resources allow.
@ -19,20 +19,31 @@ This repository is proposed to be called jobcomm (jobcommunity on top of obcommu
This means that you will see the package recipe (PKGBUILD and remaining help files), you clone the repository and build the package yourself, as described in [[how to build pkgs in joborun wiki|howto.md]].
The pacman.conf entry for this would be:
** Note: Since we have switched from OSDN mirrors to Sourceforge single catch-all repository the following is replaced.
<pre>
[jobcomm]
#Server = file:///var/cache/jobcomm/
Server = https://osdn.net/projects/joborun/storage/jobcomm/
Include = /etc/pacman.d/mirrorlist-jobo
</pre>
where jobo-mirror will contain repositories such as:
<pre>
Server = http://downloads.sourceforge.net/joborun/r
</pre>
and it should be placed right below the jobextra entry and above ArchLinux [testing].
Unfortunately you will have to flip between local and osdn repository every now and then to check for additions/upgrades. BUT!! Your main installation can have the osdn repository address and your buildbot [jobbot] chroot can have the local repository. So when your daily runner system sees and upgrade from osdn, you go to your jobbot, build the package, place it in /var/cache/jobcomm/ and then link the package to /var/cache/pacman/pkg/ so your installation can find it and install it.
But here there is a glitch, I can make the osdn database from the packages built on my machine, and the checksums of the database will reject yours, so either just use <pre>sudo pacman -U /var/cache/jobcomm/pkgname-pkver-*pkg.tar.xz</pre> or you have to be flipping /etc/pacman.conf from local to remote repository.
But here there is a glitch, I can make the osdn database from the packages built on my machine, and the checksums of the database will reject yours, so either just use
<pre> sudo pacman -U /var/cache/jobcomm/pkgname-pkver-*pkg.tar.lz <pre>
or you have to be flipping /etc/pacman.conf from local to remote repository.
Generally the reason for this is that some large packages take too much space in the limited osdn server, and many of those are binary packages that makepkg just explodes in src/ and stages them in pkg/ and repacks them into a package. So basically it is time to download a set of binaries (most browsers take hours to compile so most get the binary from source which is usually is produced in debian .deb format or fedora .rpm) uncompress and explode the archive, rearrange (seconds or less), throw back into a tar archive and compress, which in turn you install by uncompressing and exploding the tar. You might as well do this on your own machine than adding a middleman, on an already risky business of downloading and installing binaries.
@ -50,7 +61,7 @@ Alternatively you can fork jobcomm into 2 repositories, your own local repositor
<pre>
[jobcomm]
Server = https://osdn.net/projects/joborun/storage/jobcomm/
Server = http://downloads.sourceforge.net/joborun/r
[jobcomm0]
Server = file:///var/cache/jobcomm0/