On Wed, Aug 14, 2013 at 05:42:29AM +0200, Ralf Corsepius wrote:
> On 08/13/2013 09:27 PM, Till Maas wrote:
>
>> My suggestion would be:
>>
>> - Packages must have a URL tag
>> - If possible, the URL should be valid
>> - If the package is completely created by Fedora, use
>>
https://fedoraproject.org
>> - If there is no upstream web page, use the Source URL, or, if the web
>> server allows directory listings, specify the directory of the Source
>> URL
>> - If the original URL does not work, try an
archive.org one and add a
>> comment to the SPEC explaining when it was noticed that the URL does
>> not work
>> - If
archive.org does not work, use the last known URL and add a comment
>>
>> An additional hack would be to add an achor tag to URLs that are known
>> to not work anymore, such as the following:
>>
>> "http://example.com/#Fedora:+does+not+work,+no+new+URL+known"
>
> -1
>
> I don't see how the effect would be different from not having an URL
> tag, except that your proposal causes more bureaucracy.
If I do "rpm -qi foo" on a package which this kind of URL it is directly
clear where the package came from and that it is not necessary to file a
bug about a bad URL. If the URL is missing, one would need to checkout
the SPEC and see which Sources are used and maybe file a bug report or
if the URL is there but broken, one might file an unnecessary bug
report. Therefore not having a URL tag leads to more work than just
adjusting the URL tag.
The effect of whether a package carries no URL:-tag or a
"http:NoURL"-tag is equivalent.
=> Forcing packagers to use a "http:NoURL" to me qualifies as silly
bureaucracy.
Whether a package should carry a URL: pointing to
to me also is of very
little practial use. If you want this to be of practical use, there
should be a per-package URL, but ... this also means bureaucracy.
Ralf