BzrEclipse and CPU load

[Guillermo Gonzalez]

Hi,

On Sun, Nov 30, 2008 at 11:25 PM, Robert Collins
<robertc at robertcollins.net> wrote:
> On Sun, 2008-11-30 at 00:51 +0100, Thomas Manson wrote:
>>
>> I've the impression that each time you modify a file, it rescan the
>> whole project (why would it take so long and so much CPU)
>
> For modest size projects (< 5000 files or so if I remember the stats
> right), just starting bzr from scratch takes up more time than actually
> doing the status operation. I thought that eclipse kept a bzr process
> around to avoid this.
>
> -Rob
> --
> GPG key available at: <http://www.robertcollins.net/keys.txt>.
>

Robert, you'r right.
bzr-eclipse keep a bzr process around and use the xmlrpc service.

As I mentioned, it's a bug ini bzr-eclipse. If the option "use text
decorators" is enabled, bzr-eclipse 'll query the last revision of
each file in the project, and the current implementation is far from
be performance wise nor good enough.

My first approach was to minimize updates and use a cache in the java
side. Now I'm thinking that mybe makes sense to implement a helper
xmlrpc call to do all the heavy lifting, and just return a list of
tuples like (file, last_revision)

Cheers,

--
Guillermo
<http://launchpad.net/~verterok>