[Bug 17145] 'wrong version of assembly manifest', ok, but which is the right one?

[wine-bugs at winehq.org]

http://bugs.winehq.org/show_bug.cgi?id=17145





--- Comment #5 from John Doe <gotterdammerung at web.de>  2009-01-27 17:14:02 ---
(In reply to comment #4)
> You could also have a look at this patch
> http://bugs.winehq.org/attachment.cgi?id=9618116 from bug 9717117 . It probably needs
> a test case to verify the behaviour (I don't know if it is correct).
> 

I thought about this behaviour (to include the 'fix' from bug 9717117):
inform the user if the version does not match, but only fail on major.minor
mismatch, ignoring build.revision.

But then i saw how _shared_ manifest versions are handled:
fail only on lower build or, with equal build on lower revision.
That's a bit nicer, i think.

> It probably needs a test case to verify the behaviour (I don't know if it is correct).

Agreed. I don't think it is correct. Depends on what modifications are allowed
beween builds & revisions.
But if Gothic3 runs with it, it can't be that wrong. ;)


The question is: if we handle versions too loosely, where and when do we fail?
IMHO it is best to be strict: Fail fast and let the user get the right version.
Thats what the guys that tried Gothic3 would have done, if they only knew which
one was the right version. To ignore build.revision looks to me like an ugly
hack ( matching the original Gothic3 development :-> ).


-- 
Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email
Do not reply to this email, post in Bugzilla using the
above URL to reply.
------- You are receiving this mail because: -------
You are watching all bug changes.