[antlr-interest] ANTLR 3.0ea2 and ANTLRWorks 1.0ea2 released!
Terence Parr
parrt at cs.usfca.edu
Mon Jun 13 11:34:51 PDT 2005
On Jun 13, 2005, at 9:00 AM, Jim Idle wrote:
> I have worked out why the color coding seems to be broken: a
> careful examination of how it breaks, and a sip of Bridgeport IPA,
> showed that it is out by 1 character position for each line. This
> is because it is only parsing correctly when the newline sequence
> is the UNIX \n only (and, I imagine, the MAC \r). Files that came
> from some Windows editor will usually have \r\n and the syntax
> highlighting (and the editor pane in general) are not dealing with
> this.
Good catch!!! yeah :)
> Personally, I found my use of the tool to be greatly enhanced once
> this worked properly. The only thing I could do with now is a bit
> more documentation on what is syntactically different in this pre-
> release. Once I found the 3.0 demo stuff I made strides forward.
> Though it is pretty reasonable to expect that if you are using the
> pre-release you will have to wing it a bit.
Yeah, sorry about that...i'm scrambling to finish up the damn thing.
The examples do help, right?
> My only fear (hoping not to start a religious war) is that
> graphical tools built in Java always seem to crash, hang, have slow
> GUIs and so on (at least on windows, and though it would be easy to
> blame Windows, I don’t think that is the issue). At the moment this
> tool suffers form the same issues – though I am loathe to offer any
> judgment at this stage in the game – I would prefer to contribute
> to its betterment.
Java guis are indeed slow, but portable...sort of. ;) As for
crashing and hanging, I find this tool not too bad for early access
release :)
> I very much like the syntax diagrams, though I have not decided if
> they are just a cool gimmick for the programmer or not yet.
They are like the "dashboard" on Mac OS X Tiger...cool, but not
always useful to experienced developers. OTOH, they are REALLY great
for tracking down ambiguities...for this, they are really spectacular :)
> They certainly will aid in documentation enormously. Perhaps we
> could have a mode that could export all the syntax diagrams at once
> (sorry if it is there and I missed it)?
Use -nfa or -dfa to get something like that from antlr itself. I
think Jean will have something to print / generate PDF for these
puppies out later.
> Some suggestions (not criticisms):
>
> I would suggest that errors from parsing the grammar appear by
> “check grammar”, or “Build”, in their own pane, instead of the Java
> console, and be able to double click them to locate the error in
> the source code.
> Don’t prompt for output locations all the time, but allow a change
> of ‘properties’ if the programmer wants to change the output
> location after the first run.
>
>
> In light of the above, perhaps have a project file that hold such
> things. Borrowing from some of the good things in like Visual
> Studio .Net 2003 (err, I mean the freeware tools such as Eclipse),
> is not a bad idea. I realize of course that there was quite an
> effort to get to this point first and such things tend to get
> filled in afterwards.
Thanks a lot for the feedback! Good stuff.
Ter
--
CS Professor & Grad Director, University of San Francisco
Creator, ANTLR Parser Generator, http://www.antlr.org
Cofounder, http://www.jguru.com
More information about the antlr-interest
mailing list