On Wed, Jun 10, 2020 at 05:47:54PM +0200, Thomas Schwinge wrote:
> The automated system runs once a day. I guess it's generally deemed
> acceptable if the source code state (commits) and the 'ChangeLog' files
> state is out of sync for one day. There surely must be some mechanism in
> place to make sure that actual GCC tarball etc. releases do have complete
> 'ChangeLog' files state.
Martin has provided a way to generate patches for all the standard branches
instead of checking them in.
I think it would be really much better to have a way to ask either for a
patch on stdout for the HEAD (i.e. only on that branch find last DATESTAMP
bump and generate ChangeLog entries since then) or just apply the changes
to the ChangeLog directly.
At least that is something I'd prefer for gcc_release and e.g. the distro
snapshots.
For the topic branches, there is the additional problem that it is
(semi)-regularly synced from mainline or some release branch and thus is
getting the DATESTAMP update commits which are now normally used as the
anchors for ChangeLog generation. The commits that are not from the tracked
branch can be intermixed with the inherited commits though and so the
question is what should we generate and into which files.
Jakub
