| Author |
 Topic  |
|
|
LarryLeonard
Tomato Guru
USA
1041 Posts |
 Posted - Jan 30 2004 : 10:19:07 AM
|
The VAX architecture page (http://www.wholetomato.com/x/products/architecture.html) says:
quote:
Fortunately, the better assistance in Visual Assist X comes without a price in performance. Our specially-designed parsers are low-priority, background threads yielding no impact on the responsiveness of your IDE. The threads awaken typically during idle time and go silent when you start typing. And since these parsers do not generate object code, they consume far fewer resource than traditional parsers.
I have a vague complaint that I'm not sure how to put into words. Maybe I'm just using an underpowered machine (P4, 1.7GHz, 512 Meg), but most of the time I feel like VAX's parsing is way, way behind me. I very often - too often - have to click the reparse button. When I paste in new code from another file, it takes it a long time (a relative term) to stop marking things with red underlines. It feels like it's not running during idle time at all, or not soon enough, or not at a high-enough priority, or something. (Told ya it was vague.)
Could you talk a little about how the "thread awakening" is suppossed to work? Could there be a user-settable "priority" slider or something, so I could tell the parser how many cycles I want it to use? Am I the only one that wouldn't mind giving up a little IDE responsiveness for quicker parsing (or at least, having the ability via the slider to specify that tradeoff for myself)?
Also, would it be possible (just in the Beta) to write "Starting parsing" and "Finished parsing" to the status bar (or something)? Just a thought... Thanks.
|
Edited by - LarryLeonard on Jan 30 2004 10:21:08 AM |
|
|
willdean
Tomato Guru
134 Posts |
Posted - Jan 30 2004 : 11:29:33 AM
|
I too support the idea that they should remove this disingenuous paragraph. They're fighting a battle between lazy-parsing latency and nasty jerky interruption, and have found a middle ground which is probably slightly irritating on both counts.
My personal opinion is that over-lazy parsing (but as you say, it often completley breaks after a paste and doesn't seem to parse at all, even with r-p) is a lot better than blocking stuff which interrupts the flow of my movements around the code. I'm currently using a very powerful machine, and still don't think that it's very smooth. Perhaps you and I would chose different points on your slider - particularly if there was a reliable reparse function.
I second the status-bar request - it would be nice to at least think something was happening while I'm staring at a screen full of uncoloured text.
|
 |
|
|
support
Whole Tomato Software
5566 Posts |
Posted - Jan 30 2004 : 11:44:19 AM
|
There was a bug that prevented VA from parsing code as it was pasted.
Will be fixed in build 1215. |
Whole Tomato Software, Inc. |
|
|
|
Uniwares
Tomato Guru
Portugal
2321 Posts |
Posted - Jan 30 2004 : 3:25:10 PM
|
| A related question is here: can we update the Scintilla dll or are we stuck with the provided one? |
|
|
|
support
Whole Tomato Software
5566 Posts |
Posted - Jan 30 2004 : 8:41:53 PM
|
We use Scintilla only for the tiny editor appearing when you open our options dialog to edit Autotext and Code Templates.
As for replacing the Scintilla DLL, you can try. Your only possible improvement is in editing of templates. You should probably forgo the exercise and edit templates with Notepad.
While on the topic, we plan to replace the template editor with a grid control. Maybe this year. |
Whole Tomato Software, Inc. |
Edited by - support on Jan 30 2004 8:44:28 PM |
|
|
|
Uniwares
Tomato Guru
Portugal
2321 Posts |
Posted - Jan 31 2004 : 3:15:30 PM
|
| Right, was just curious... |
|
|
|
LarryLeonard
Tomato Guru
USA
1041 Posts |
Posted - Jan 31 2004 : 8:25:24 PM
|
So, annnnnnyway, what about my original issues? ...
|
|
|
|
support
Whole Tomato Software
5566 Posts |
Posted - Feb 02 2004 : 8:55:17 PM
|
| Parsing is done after every paste in build 1215. |
Whole Tomato Software, Inc. |
|
|
| |
Topic |
|
Topic Locked
