Alec, On Monday, July 1, 2024 4:19:22 PM MST Alec Leamas wrote: > Hi again > > On 02/07/2024 01:13, Scott Kitterman wrote: > > On Monday, July 1, 2024 7:07:16 PM EDT Alec Leamas wrote: > >> On 02/07/2024 00:54, Scott Kitterman wrote: > >>> On Monday, July 1, 2024 6:46:06 PM EDT Alec Leamas wrote: > >>>> If you switch hats for a moment: have you any advice for upstream in > >>>> this situation? > >>> > >>> 8763.5.10 > >> > >> Yes, I have had a similar idea using 10000 instead of 8763 to make it > >> stand out less. In my eyes, this is worse and will lead to that the > >> package versions does not match the "public" version like 5.10.2. > >> > >> But if the list agrees that this is the correct solution so be it. To be > >> honest, it might be a hard sell upstream. > >> > >>> Next build is: > >>> > >>> 8763.5.10~8764 > >> > >> Why? > >> > >> --alec > > > > Because the '~' means less than. It's a way to add the build number to the > > interim versions in the future without causing the same problem again. I > > guess it should have been 8763.5.11~8764, if 5.11 is the next 'real' > > version. > > There is absolutely no need of build numbers in the version, it's just a > sad legacy. > > Let's drop this subthread, keeping eyes on the ball: what is a sane version? > > --a
If upstream wants to fix this problem, they could just make their next release version 9000, with the one after that either being 9001 or 9000.1. Or, possibly, they could encourage everyone to uninstall the PPA package before installing an official one. For example, release a new package to their PPA that displays a message encouraging everyone to uninstall the PPA package, remove the PPA from their list of repositories, and *then* install the official one. As a general rule, I wouldn’t expect a user to keep a PPA package installed when switching to an official package. There is generally no guarantee that upgrading from a PPA package to an official one will work without errors. Or, once the official package had entered the system, they could instruct users to remove the PPA from their list of repositories and then perform a downgrade. All of that being said, Debian could use an epoch to fix the problem. Having an epoch on a package isn’t the worst thing that has ever happened. -- Soren Stoutner so...@debian.org
signature.asc
Description: This is a digitally signed message part.