> So, after thinking about this for much longer than I'd hoped to, I
> believe my dissatisfaction with the ways this was implemented can be
> described briefly like so:
>
> * Properties are getting far too specialized a level of treatment.
> They are an artifact of a modified versioned object, no more or
> less so than a file's contents.
>
> * While I appreciate the attempt at consistency in diff output, in
> this one case it's problematic because patch won't gracefully
> ignore propdiffs that it ultimately can't apply anywhere.
>
> * The configury is just far too complex for such a dinky little
> feature. I don't perceive a true need for toggling propdiff
> display based on the action performed on the property, but instead
> think of a file's properties as artifacts of that file,
> and therefore the display thereof as being toggled in the same
> fashion as the display of the file's contents/diff.
>
> I guess I'd like to see mailer.py retain its paths-as-primary-keys
> approach, with the artifacts of paths (content/content-diffs, repository
> viewer urls, and properties/propdiffs) all treated equally.
>
> I'll try to make a solid proposal in a follow-up mail, probably by
> including a sample commit mail annotated with the toggleable regions
> (and noting specifically the toggles).
I feel terrible about this, but a SOLID YEAR went by without me actually
making time to make this follow-up proposal. That said, I still feel
strongly about the code as committed. I realized just this morning that
your changes, Mathias, were still present on the trunk version of mailer.py.
I hope you can forgive me for doing so, but I just committed the reversion
of them (in their entirety).
Again, I like the idea of better diff displays. But I really disliked the
approach taken (for the reasons above), enough so to not want to ship it in
Subversion 1.5. (If it's not obvious, I'm soft-pedaling around the code
"-1" stamp of disapproval here.)
I will open an issue in the issue tracker with your changes attached as a
patch. I know they live in VC history, but the issue will remind us to ...
well, not forget them.
--
C. Michael Pilato <cmpilato@collab.net>
CollabNet <> www.collab.net <> Distributed Development On Demand
Received on Tue Sep 25 16:32:13 2007