[PATCH] convert: add --authormapsuffix to append set text toauthor names (issue4171))

[Sean Farley]

Augie Fackler <raf at durin42.com> writes:

> On Sat, Feb 15, 2014 at 11:50:03AM +1100, Lawrence Stewart wrote:
>> On 02/14/14 14:57, Lawrence Stewart wrote:
>> > Hi Mads,
>> >
>> > On 02/14/14 12:52, Mads Kiilerich wrote:
>> [snip]
>> >>
>> >> You mention the lack of a more complete templating like feature. How
>> >> could it look like? Could we aim at that instead?
>> >
>> > We certainly could and probably should. Only reason I proposed this
>> > patch is that it exists now, works as advertised and I've been using it
>> > during my setup of http://hg-beta.freebsd.org/ (currently offline for
>> > another few hours while I rerun the conversion with my --authormapsuffix
>> > patch so that authors are mapped to @FreeBSD.org so that downstream
>> > consumers of the repo can better differentiate from changes made by
>> > upstream authors vs local authors).
>>
>> FYI, http://hg-beta.freebsd.org/ is back up if anyone is interested to
>> take a look. It is being instantiated as a read-only seed repository to
>> facilitate non-blessed committers and organisations developing
>> FreeBSD-derived products to be able to more easily track and hack
>> FreeBSD. I am putting together corresponding documentation to explain
>> how to do useful things with it, which might be interesting to get some
>> Mercurial community feedback on at some stage.
>>
>> The FreeBSD svn source repo represents something of an interesting
>> conversion challenge, with the partially converted repo (trunk and
>> stable branches only currently, more branches will likely follow later)
>> coming in at ~240k commits, ~1GB size on disk.
>
> Have you looked at hgsubversion at all? The facebook guys have done a
> lot of work that might make this more straightforward.
>
>>
>> As an aside, being able to detect MFCes (cherry-pick merges from the
>> head/trunk branch back to stable branches) and add appropriate ancestral
>> links during the conversion would be a nice outcome of a more advanced
>> snippets/templating based arrangement. As far as I know, it's currently
>> impossible to achieve this on-the-fly when doing incremental conversion
>> runs against the svn source of truth.
>
> Does svn:mergeinfo get recorded on these merges? If so, it might be possible.

It does and it's how I implemented the experimental merge history with
hgsubversion that I'm in the process of patch bombing :-)

Though, if someone could tell me how to distinguish between a
cherry-pick and merge, that would be great.