|
Bugzilla – Full Text Bug Listing |
| Summary: | YOU selects wrong package for installation | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE Linux 10.1 | Reporter: | Wolfgang Rosenauer <wolfgang.rosenauer> |
| Component: | YaST2 | Assignee: | Klaus Kämpf <kkaempf> |
| Status: | RESOLVED INVALID | QA Contact: | Stanislav Visnovsky <visnov> |
| Severity: | Normal | ||
| Priority: | P5 - None | CC: | andreas.hanke, hmuelle, meissner, suse-beta |
| Version: | Final | ||
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | SuSE Linux 10.1 | ||
| See Also: | https://fate.suse.com/300899 | ||
| Whiteboard: | |||
| Found By: | Other | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
| Attachments: | screenshot | ||
|
Description
Wolfgang Rosenauer
2006-07-31 17:52:29 UTC
Created attachment 94850 [details]
screenshot
This screenshot shows different problems:
- the trash icons
- the package from the inofficial repo is chosen
while the patch should have fixed packages referenced
which should be chosen
- the wrong architecture is chosen in the UI (x86_64)
while the installed package is i586
I guess but haven't tested right now that when I confirm this automatic settings that seamonkey will be updated to the 1.0.3 version but with i586 package.
(That's what happened with Firefox some minutes ago)
and BTW: is the estimated size of the download correct with 0 B ? I think this is already known because very obvious. The patch only requires a minimal version, not an exact version. So if a newer package is available, it will be installed. If you want a specific version, the patch must be adapted. What about the other bugs in the UI. Should I file extra bugs on it? Looking at the patch xml file I see direct references to specific packages with sizes of packages and so on. So I would expect that a patch always has specific packages with it. If it doesn't have I really think it should have or we should provide patches which do this. I really don't think the current behaviour is the "right" one. If that's just a matter how a patch is configured I recommend to change that to require the exact version. (In reply to comment #4) > So I would expect that a patch always has specific > packages with it. If it doesn't have I really think it should have or we > should provide patches which do this. Be careful: This is possible, but extra care has to be taken. With this approach, if a package is updated multiple times (which will definitely be the case for Firefox), the subsequent patches must explicitly obsolete each other. Otherwise it will result in conflicts/failed installations like bug 196423. > Be careful: This is possible, but extra care has to be taken. With this
> approach, if a package is updated multiple times (which will definitely be the
> case for Firefox), the subsequent patches must explicitly obsolete each other.
This is the reason exactly to have ">=" relation and a "higher version-release" is normally the better package.
Imagine you have "==" releations in patches and multiable repositories from various sources. There will be situation where you have conflicts cause there are newer (better packages) in the various sources but the resolver can't use them because a fixed version is required in a certain patch.
It is hard to impossible to define "best package" cause various people will have various preferences, but relying on versions is the best solution, isn't it?
I think "selecting wrong package" is invalid, for other bugs (UI) please open a new bug.
|