-
Notifications
You must be signed in to change notification settings - Fork 0
Conversation
I don't like the idea of maintaining files on the distmaster via spurious commands. There are far too many files already 'distmaster-only'. We lose information about where the files come from ( Why not just update the |
That a (soon outdated) snapshot of the ID database is shipped in the source archive is just a convenience, but gets in the way for distributions in my opinion. I do not see the point of rebuilding source files just to get updated IDs. One could create an extra package with just one file as Debian does it. But, in practice for MarIuX, once the database is taken out and during installation newly created, the next time we want to update the ID database, we will see, that it’s not packaged, and quickly find out, that just one command needs to be executed on the distmaster. No package churn, no merge/pull requests, and so on. |
Perhaps build a database of all files, which are not installed by a bee package and teach "bee query" to show the sources of these files, too? |
Lets see how often this happens:
Six times in ten years.
On every update, the source release version and |
Why ? As shown before this just happened 6 times in TEN years.
When will this happen ? In 1.6 Years ? I would think the source would changed and needed to be updated until then, too.
How do we find out it's not packaged ? What commands would an admin have to enter to find out it's not packaged ? It's much easier to find it is packaged, than the opposite.
How do I find out what command is linke to create this file ? For me, it's
These changes render the update command useless. You have to touch the source.
Questioning merge/pulls ? |
Who updates the database ? A cron job ? Should it track changes to these files ? Keep some history ? Where do we put this database. Would it honor Distfile-excludes ? Then on which host should it be created ? Why should we make already solved problems more complicated ? There is a file somewhere, it belongs to a bee file and is mentioned in |
Thats more a general idea, not strictly related to this PR. We already have many many files on the distmaster which do not originate from bee packages. Some do have other defined sources (mxtools), some are created by runtime or boot tools (/etc/exports), some where manually created in ancient times and are not tracked at all (/etc/resolv.conf). So a database which enumerate the source for every file might be nice. No, it should not be build by cron and document the moving status quo which exists on the filesystem anyway. Registering a file would be done by the tool which installs it, or by an administrator. We could create a notification if unregistered files appear. |
|
I'm not questioning a
I will hardly accept any new Distfile exception. You found a solution, even two, so why not stick with it. If you think updating the bee package is too much work, feel free to delegate this work to me. I'm happy to do this for you with every update of pciids. Count on me. You still owe me answers to
All this hassle because you don't want to 'quickfix' use |
> 3. Getting new IDs > ~~~~~~~~~~~~~~~~~~~ > The database of PCI IDs (the pci.ids file) gets out of date much faster > than I release new versions of this package, so it is maintained separately. > > It lives at http://pci-ids.ucw.cz/, where you can browse the database, > download the most recent pci.ids file (e.g., by running the update-ids utility) > and also submit new entries. > > Alternatively, you can use `lspci -q' to query the central database > for new entries via network. Therefore, maintain `/usr/share/hwdata/pci.ids` directly on the distmaster.
A manual page for `pci.ids` is added. [1]: https://github.com/pciutils/pciutils/releases
44f9e2f
to
779a415
Compare
What is the dependency? The system file is there.
It’s only updated, if you run a command. So hardly belongs into
Isn’t that for things, how we configure MarIuX?
Too much hassle.
This file is supposed to be disted.
Because I will forget about those.
Thank you for the offer. I’ll guess I’ll take you up on that.
Ditto.
Too much work, and did not work out. You need to change the bee file, to additionally get the latest version. So, the package is not reproducible, as it can download newer IDs, if you run it later.
You cannot update that file. But no downside compared to outdated.
What location? You still have at least the IDs of the last release.
Format of what? Very theoretical question. You go back to the latest working version from the backup. But they will maintain compatibility.
Yeah. Run Several of your concerns (change format, server not reachable) apply to shipping it in the bee package too. My solution is not worse. |
I put the file back into the package. |
Peter volunteered to keep the PCI IDs up-to-date. Increment revision to 1.
0296c6f
to
2fbe1d0
Compare
Thanks. Feel free to merge, drop me a message so I'll rebuild on db update. |
@wwwutz volunteered to keep the PCI IDs up-to-date, so keep the database in the package.
Tested on rabammel and amaru. On amaru, the Intel graphics device is correctly named now.