[pedevel] Re: keywords files.

  • From: Oliver Tappe <pedevel@xxxxxxxxxxxxxxx>
  • To: pedevel@xxxxxxxxxxxxx
  • Date: Mon, 25 Apr 2005 18:22:51 +0200

On 2005-04-24 at 23:12:55 [+0200], BiPolar wrote:
> Oliver Tappe wrote:
> 
> > > Is that feature lost for good, or its just a side effect of recent 
> > > changes?
> > 
> > Well, you could call it a side effect, but I'd call it a bug >:oP I was 
> > simply a bit too forceful when replacing the keywords-related code. Now 
> > the original behaviour has been re-added, i.e. Pe automatically 
> > generates a specific text-file for the keywords of each language. The 
> > only difference is now, that these files are only generated when a 
> > language is actually being used (aka: on-the-fly).
> 
> Good, so no more big delays upon fisrt pe start after a clean install, 
> right?

Correct.

> I must add that reducing the size of the keyword.cpp file made pe load 
> faster on my (slow?) system (K7@900), specially for big C/C++ files (> 
> 100 KB). It wasn't "slow", but not as quick as I became accustomed to.

Ah, ok. I will look into the performance of keyword-highlighting a bit 
deeper tonight.

> I've faced a little annoyance recently: Key-bindings do not work for some 
> of the extension addons.
> 
> I've tried erasing the old keybinding setting file and adding the keys 
> again, but that didn't helped.
> 
> For example: if I bind "F1" to BeBookFetch or ROT13, nothing happens when 
> I press F1 (but the addons *do* work if I invoke them from the menu). 
> But, if I bind F1 to, for example, HeaderGuard, it works as expected :-/
> 
> Can anyone else confirm this behavior? (maybe its just my bad luck)

Yes, I can confirm that. It was due to my recent change of how Pe handles 
extension-shortcuts. It chose to use the index of the extension in the 
submenu as key for shortcuts, which had the (unpleasant, in my eyes) 
side-effect that adding / removing an extension could mess up the 
shortcuts. I fixed that by replacing the menu-index with a 16-bit hash. 
However, when doing so, I introduced a signedness bug that caused 
extensions that happen to have a hash-value with the topmost bit set to not 
react to their shortcuts. That's fixed now, of course >;o)

Oscar, thanks for the heads up (again).

cheers,
        Oliver

Other related posts: