WIP: new VCL scheduler feature branch

[Noel Grandin]

On 2016/09/19 2:27 PM, Jan-Marek Glogowski wrote:
>
>> Some random low priority comments:
>>
>> In "Just walk the task list once per timeout"
>> https://cgit.freedesktop.org/libreoffice/core/commit/?h=feature/new-vcl-scheduler&id=435e21fe0330436e76e5e053d5d5d94df734a554
>> The evaluate_entry label doesn't make the code any easier to read.
>
> So you're suggestion is? Introduce an other level of indention for the
> if? The while loop is short enough and the goto labels are just used in
> there. If you think it improves readability, I can change it.
>

No, just inline that single line of code at the evaluate_entry label and then jump to the exit label.
But that's just my taste, if you disagree, feel free to ignore.

>> In "Reorganize Scheduler priority classes"
>> https://cgit.freedesktop.org/libreoffice/core/commit/?h=feature/new-vcl-scheduler&id=5e4361e84607fc6d7623b31630505da7c934b945
>>  (1) you haven't used a consistent mapping from old to new - sometimes
>> LOW maps to HIGH_IDLE, sometimes HIGHEST maps to HIGH_IDLE. Perhaps
>> there is a reason for this?
>
> From naming and code reading I tried to guess, if the idle should be
> processed before drawing. As I wrote I would like to get a comment or
> annotation for all non-default priorities, i.e. all calls to "SetPriority".
>
> I have no idea, if my guesses are correct, just that mail merge is much
> faster now and LO is still usable while the mail merge runs. Actually
> you can see the slowdown of mail merge when doing stuff in LO :-)
>

OK, just checking, because from the commit message it looks like a mechanical change, but then the mapping was not 
consistent.

>> (2) I would call HIGH_IDLE, either PRE_RESIZE or BEFORE_RESIZE, because
>> it's not any kind of IDLE anymore. I would call DEFAULT_IDLE just IDLE.
>
> I just took the priority names from glib. I would keep the DEFAULT_IDLE,
> or rename DEFAULT to TIMER? OTOH the Scheduler class sets the default
> priority...
>

Ah, then probably leave them alone, if they map to something we already use.