Doing away with changelogs
lxs at MIT.EDU
Wed Apr 19 13:42:46 EDT 2006
On Apr 18, 2006, at 9:07 PM, Henry B. Hotz wrote:
> On Apr 18, 2006, at 7:53 AM, krbdev-request at mit.edu wrote:
>> I suspect this need would be fulfilled by just improving our release
>> notes so that they list all the bugs fixed and the svn commit lines
>> to show what files were impacted by each change. Associating each
>> change with a bug number would in fact be an improvement over the
>> Changelogs since you'd know what bug needs to be reopened/referenced
>> once you discover which change broken things.
> Sounds like you may just be shifting the "inaccessible off-line
> repository" problem from svn to rt. Not clear.
> The point is that the ChangeLog file ought to be complete of itself.
> If you want to make it more complete by pulling in info from your
> bugs database in addition to the change notes themselves then I'm all
> for it (but I'm not writing the software to do it).
The changelogs are already incomplete in this manner. Each entry in
a changelog already has an associated RT bug which may contain more
information than what is in the changelog. For example, if a user
files an RT bug about a bizarre GUI behavior that turns out to be a
side effect of flow control in a low-level API, the changelog entry
for that change may only describe how the flow control changed and
not the actual bug fixed by it.
With the autogenerated script one could choose to include all
information in the RT bugs, not just a description of the code
change. Which is why I said it could (depending on how it's
implemented) be an improvement over the existing changelogs.
Alexandra Ellwood <lxs at mit.edu>
MIT Kerberos Development Team
More information about the krbdev