[Ticket#: 2006040510000641] Re: [OT] Articles to read
J. Landman Gay
jacque at hyperactivesw.com
Thu Apr 6 18:53:38 EDT 2006
David Burgun wrote:
> Just look at the Script Editor! There are at least 10 bugs that I come
> across every day, and for 2+ years???!!!!
Did you put them into bugzilla? If you didn't, the team won't always
know there is a problem.
>
> For instance, straight off the top of my head:
>
> 1. I open a script and do nothing to the script, I then close the
> script and it gives me the "Do you want to save?" dialog.
I'm not sure this is really a bug. It is more a by-product of one of the
features. Revolution saves not only your script, but also the location
of the insertion point so that the next time you open the script it will
scroll automatically to where you were last working. This is a nice
feature. So, if you close the script editor without clicking anywhere,
you won't get the "save" warning. If you do click, it marks the script
as dirty, stores the current cursor location, and asks if you want to
save it.
Once you understand what's happening, it may not seem so terrible any more.
> 2. The script colorization doesn't work consistently.
> 3. Sometimes the script editor/debugger loses it's mind and when you
> click on the error it opens the *same* script in another window. If you
> then edit this window and close it, then notice that the other window
> is still open and close that one, you lose your changes!
> 4. The debugger sometimes just plain refuses to breakpoint, even when
> you insert a "breakpoint" command into the script.
> 5. The Menu Builder tool is really flakey.
These are mostly valid and I've also seen some of them. I can't comment
on colorization because I abhor it and always turn it off. (It also
increases the size of your stack on disk because Rev must store a
duplicate htmltext copy.) I've seen the double script window problem
only a couple of times, but it needs to be reported. If you have a stack
that reproduces the problem, I hope you will submit it so it can be
fixed. I've never had any trouble at all with the menu builder, but if
you can define the problem more concisely than "flakey" the team can
look at it. It has always worked okay for me.
Number 4, refusing to break, happens only under very particular
circumstances. I've been able to reproduce it twice, reported it twice,
and both times it was immediately fixed. In each case, it was a problem
with a library or a backscript -- once with libURL and another time with
the way the IDE was parsing file paths. I think it happens when a
library is in use in a stack that is not open -- or when a backscript
has been inserted -- and the library has a bug. The debugger will not
trace into that script. The engine has two choices at this point; the
first is to go into an infinite loop (that's what the MetaCard IDE does)
and the other choice is to simply not break at all. Rev does not break
at all. This is infinitely superior to a loop that you can't get out of.
When I used to hit this problem in the MetaCard IDE I had to force-quit
without the opportunity to save my work.
Careful debugging of your own library scripts can help, if that is where
the problem is. If the problem is with one of the backscripts or
libraries that Rev employs, then *report it.* I just did a couple of
searches in Bugzilla, and did not find any bug reports submitted under
either your name or email address. You can't expect fixes to things that
aren't in the bug database.
Note also that the team uses Rev and all its features daily to develop
Rev's IDE, among other things. Therefore, if you are having a problem it
is likely localized to either your setup or your work style. They need
to know. For example, one reason my copy of the debugger would not break
was because of a file path issue that existed only on my hard drive. No
one else could reproduce it; I finally figured it out when I changed the
name of a folder. Then I reported it.
--
Jacqueline Landman Gay | jacque at hyperactivesw.com
HyperActive Software | http://www.hyperactivesw.com
More information about the use-livecode
mailing list