[josm-dev] Commit message not empty

[Matt Amos]

On Fri, Apr 24, 2009 at 11:02 PM, Frederik Ramm <frederik at remote.org> wrote:
> Matt Amos wrote:
>>
>> On Fri, Apr 24, 2009 at 6:05 PM, Richard Fairhurst <richard at systemed.net>
>> wrote:
>>>
>>> Well, in reality Potlatch can put any comment where the hell it likes
>>> because it's not hidebound by the XML API that you poor saps have to
>>> labour
>>> under...
>>
>> and, in reality, so can JOSM. the XML api doesn't require a comment
>> tag on the changeset at any point.
>
> People asked why the API does not *enforce* a commit comment if it was so
> important. My answer that Potlatch would not be able to function then. Which
> is true (if one assumes that Potlatch would use the normal API...); because
> how would you enforce a commit comment other than requiring the tag be set
> on changeset creation?

which is one of the reasons i don't like enforcing comments on
changesets. the others are the same as why we don't enforce any sort
of tagging schema - i prefer to give users and developers the chance
to surprise me by doing something cool i didn't expect ;-)

>> of course, if what you meant was that because you write the potlatch
>> api you can reach into the guts of the database and fiddle with
>> changesets after they're closed... well... :-P
>
> Oh, so we didn't implement the cryptographic anti-Potlatch-fiddling measures
> that we talked about on the hack weekend after Richard went home?

sshhh! he still doesn't know we de-prioritise potlatch api calls so it
becomes unusably slow and everyone switches to using JOSM or
merkaartor instead.

cheers,

matt