http://bugzilla.spamassassin.org/show_bug.cgi?id=3523





------- Additional Comments From [EMAIL PROTECTED]  2004-06-26 06:04 -------
> Agreed with Duncan -- "-dev" is more appropriate, since it implies 
> instability ;) 
 
k, fine with me. 
 
> I don't even think including the branch name is necessary, as I don't want 
> to get into having > 1 "dev version" available on the website.  (anyway, 
> "-x.y-dev" isn't appropriate now that we have 3-part version numbers, 
> "-3.0-latest" should be "-3.0.x-latest" or something.) 
 
That one I don't understand, but maybe I was a bit unclear :) With "branch" I 
meant the Y part in the X.Y.Z versioning scheme. So what I did was to leave 
the Z part away for "HEAD" as it isn't really correct; the nightly build is 
not yet a X.Y.Z (ie. 3.0.0), it's the current state of the X.Y branch in 
question (ie. 3.0). So I took the Z-part as a release tag which is only 
present for the final version. Could also replace it with an "x" (ie. 3.0.x), 
just having the upcoming "release" number in there doesn't sound right to me. 
 
About the > 1 dev version: I meant the case when we already started a new HEAD 
and backport stuff to a STABLE branch parallely (as we normally do). Then we'd 
have for example a 3.0-dev and a 3.1-dev (or 3.0.x-dev/3.1.x-dev) lying 
around, one for each branch. 
 
>> - If no gppsign script is found, the release will build successfully, 
>> unsigned but print a fat warning in the end.' 
> 
> -1.   Could you make that behaviour something that's enabled by an optional 
> commandline flag, or env variable, or something?   We don't want the nightly 
> scripts to suddenly stop signing the files, and nobody would notice because 
> the builds scripts carry on fine but the warning message never gets seen. 
 
Of course -- it's now just the way it always was, when no script was 
available, it went on silently. No problem making it exit with an error 
though. 
 
> OK, I see one more problem -- why not use "make tardist" and "make zipdist"?  
> I'm not sure I see the benefit of not using those. 
 
When I tested the build I had the very annoying behaviour that it rebuilt the 
whole dist again and again which took ages :-/ Maybe its a bug in my version 
of EU::MM, but in the end I just replaced it with the literal commands and 
everything was fine. 
 



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to