On Mon, 2009-08-24 at 09:32 +0000, Martin Pool wrote:
> The uses of path_content_summary in the test suite seem to be:
>
> 1- test_commit_builder, that use it as input to record_entry_contents:
> we could change them to pass in the expected (generally pretty simple)
> value instead
>
> 2- test_path_content_summary - these probably want to be updated to
> get_kind_and_executable
>
> 3- as for #2 but in test_transform
>
> I think overall this is not an inherently dangerous api, but it is also
> not one that we particularly want to encourage people to use because it
> will tend to do too much work in reading in the whole file. What is
> dangerous here is that there's no clear systematic distinction between
> when we're talking about the convenient and when about the canonical
> forms.
So, its worth noting I think, that we got significant performance
increases by path_content_summary, because it avoided path computation
multiple times inside tree; the new commit code path largely addresses
that :- we're actually rolling back a change here ;)
On Mon, 2009-08-24 at 09:32 +0000, Martin Pool wrote: summary in the test suite seem to be: builder, that use it as input to record_ entry_contents: content_ summary - these probably want to be updated to and_executable
> The uses of path_content_
>
> 1- test_commit_
> we could change them to pass in the expected (generally pretty simple)
> value instead
>
> 2- test_path_
> get_kind_
>
> 3- as for #2 but in test_transform
>
> I think overall this is not an inherently dangerous api, but it is also
> not one that we particularly want to encourage people to use because it
> will tend to do too much work in reading in the whole file. What is
> dangerous here is that there's no clear systematic distinction between
> when we're talking about the convenient and when about the canonical
> forms.
So, its worth noting I think, that we got significant performance summary, because it avoided path computation
increases by path_content_
multiple times inside tree; the new commit code path largely addresses
that :- we're actually rolling back a change here ;)
-Rob