Terr:

What I have noticed with ANTLRWorks. The first version I started with didn't 
seem to have the performance and stability problems. My current version 1.3.1 
does have them. I think I went through two semi-automatic upgrades (Two 
notifications that upgrades were available which I naturally accepted.) This 
version checks for ANTLR syntax errors with every keystroke, which I don't 
recall for my first version. So I would definitely check out recent changes.

The slowdown takes a while to appear. I think I have it more often when I am 
working with the string template generator files than with my large toy 
grammar, but I have definitely seen it with both. FWIW my suspicion is a memory 
leak, but there are other possibilities. I believe there was talk a few months 
ago of using Java's unbuffered IO as the default for ANTLR itself, which can 
have serious performance problems, but I suspect is not the cause here. Java's 
poor default hash function could also be a contributor, but that suspicion 
could just be because I am reading Valloud's text, and he has strong opinions 
on that topic.

On Feb 1, 2010, at 12:40 PM, Terence Parr wrote:

> Hi,
> 
> Jean and I are re-factoring ANTLRWorks to be much simpler in anticipation of 
> me taking over maintenance and adding features. The first thing I want to do 
> is find out why AW sometimes slows to a crawl. I can't seem to reproduce it 
> when I want to, although it was really really slow the other day before I 
> decided to debug it! ;)  
> 
> So, I need your help to figure out when it occurs so that it will give me a 
> clue about where to look in the source code.   My impression is that one of 
> the triggers seems to be when there are lots of errors. But, when I added a 
> whole bunch of grammar windows and then started randomly inserting errors all 
> over the grammar, memory usage didn't really go up and it didn't slow down at 
> all (lots other errors to the console, but that's it). I wonder though if it 
> was getting more memory available to it in the debugger than when I run it 
> from the jar file. Also, I note that sometimes undo is extremely slow, so 
> that is also a potential area to look.
> 
> My current candidates for slowdown:
> 
> * running out of memory and thrashing
> * undo mechanism is tracking too much or something
> * error mechanism somehow  tracks too much or has too much work to do
> 
> Each keystroke triggers colorizing the editor window, but Jean tells me he's 
> sure that the parser itself is okay. might be the error handling though.
> 
> Can anybody reproduce a case where you start ANTLRWorks and then do some 
> stuff in the editor window with a grammar (of any size) and it starts to get 
> really slow to type? Even if you can't send me your grammar, I can send you 
> an instrumented version of AW that will turn out some information about what 
> is going on.
> 
> Thanks! This is my highest priority for AW since I'm using it to build v4 
> ANTLR.
> 
> Thanks,
> Ter
> 
> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> Unsubscribe: 
> http://www.antlr.org/mailman/options/antlr-interest/your-email-address
> 


List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: 
http://www.antlr.org/mailman/options/antlr-interest/your-email-address

-- 
You received this message because you are subscribed to the Google Groups 
"il-antlr-interest" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/il-antlr-interest?hl=en.

Reply via email to