WebKit thumbnailer

[Jérôme Guelfucci]

On 26/05/11 23:18, Jannis Pohlmann wrote:
> On Thu, 26 May 2011 23:10:25 +0200
> Jérôme Guelfucci<jeromeg at xfce.org>  wrote:
>
>> On 26/05/11 16:08, Jannis Pohlmann wrote:
>>> Hey,
>>>
>>> Jerome, I just tested your WebKit thumbnailer plugin and it works
>>> for me. I adjusted the coding style a tiny bit and made the
>>> thumbnails be generated from based on a 1024x1024 frame instead of
>>> 1024x600 (since thumbnails are more or less supposed to have
>>> quadratic dimensions).
>>>
>>> I rebased it on master and pushed it to jeromeg/webkit-thumbnailer
>>> instead of webkit-thumbnailer.
>>>
>>> Anything else that needs to be done except maybe add a few comments
>>> here and there?
>>>
>>> Cheers,
>>> Jannis
>>
>> Hello,
>>
>> I can add more code comments, that's not an issue. I'll work on this
>> this weekend, I'll have some time.
>
> Ok, that would be cool. Just so that others have it easier when they
> try to understand what webkit-thumbnailer.c does.

Sure, no problem.

>> Re further work, the only thing I could think of is freeing some
>> memory by blanking the page loaded in the Webkit view in an idle
>> callback (when the thumbnailer plugin is not used for a while).
>
> Ah, so the page last thumbnailed stays in memory?

Yes.

Is there a way to
> measure how much memory we are talking about (typical web pages
> downloaded from the internet)?

No idea, I'll have a look at that.
>
> Performance-wise I suppose that switching back to a blank page after
> every generated thumbnail is too much. A timeout sounds more
> reasonable. But then again, tumblerd shuts down by itself if it hasn't
> been used in a while, so maybe that's good enough. So I think
> this is worth only if the average memory footprint of the pages that
> stay in memory is notably high.

That's exactly what I thought, I'll get back to you when I have some 
figures. Anyway, I don' think this is critical, so it should not be a 
blocker for entering the master branch.

Jérôme