From: Jonathan Johnson <[EMAIL PROTECTED]>
On Oct 11, 2006, at 11:16 PM, Paul Rodman wrote:
Yup. RBScript has both been fixed and completely screwed up and
useless.
Perhaps I'm missing something, but do you have feedback IDs for the
RBScript bugs that render it useless? As far as I'm aware, it's
working much better than it has in the past, and I know at least one
person who had over 15 reports open that has told me all of them are
now fixed. I can only find one open report that is new, which is that
paramarray's on the context lose the last parameter, but I can't see
how this means that RBScript is useless.
Thanks,
Jon
(sorry about the delayed reply - I'm in Digest mode)
My application uses RBScript extensively. My users are creating
sophisticated scripts that extend the application in many ways. I
prepend several thousand lines of RBScript code to their scripts in
order to get around the onerous (but understandable) restrictions of
RBScript. I worked around the couple of limitations (i.e. bugs) of
RBScript (V5.5.5) in various ways (e.g. Variants, boolean arrays).
Until recently (RB2006r2), other (non-RBScript) bugs forced me to use
5.5.5. Now the application is OK , and a slew of RBScript bugs were
fixed, as you point out. However, RBScript was completely inoperative
in the betas until just before the release of RB2006r4 and I was
unable to test more than trivial RBScripts. Sure: my sample test apps
now appear to work. However, my main app is as flakey as all heck.
With a PEF build, no scripts will run at all: I get a dialog with:
"failed assertion: RBScriptLinker.cpp line 162".
With a Macho-O build, some of the scripts run, and some just crash
the app silently. Some will run once or twice and then crash the app.
The crash logs tell me nothing.
Haven't tried Windows yet.
I haven't reported these since,
a. I only started major work on this stuff yesterday.
b. Compiles take forever compared to 5.5.5, so I tend to go take a
leak, get coffee, etc. (and then forget I was compiling).
Occasionally the IDE crashes, which doesn't help.
c. As I mentioned, there are several thousand lines of code with no
way to break on exceptions, single-step, etc. So I have to work
through the giant text file of code by a process of
divide-and-conquer.
d. I'm not willing to report anything until I can show that it isn't
my script code that's causing it (even though it's compatible with
V5.5.5). I'd rather report a problem together with a test project
that illustrates it.
e. My project is way too complex to hand over as a test case.
f. I really don't have several days to do this. Hey, who's paying who here? :^)
REALbasic is a tremendous piece of work, and a godsend for Mac
cross-platform work, but it's clear that it's not tested to the level
it should be. Understandable, considering the expense of
weapons-grade in-house testing. However, I suspect that those of use
who make a living (sic) using RB would much, much, ..., much rather
see more effort put into getting a bug-free release than new features
(UB support aside).
Another question is: when are the RB developers going to start
"eating their own dog food"? When they start using RB2006 on a
full-time basis, I'll bet all those "minor" annoyances (like the
insertion point being out by a pixel or so, etc.) will vanish almost
overnight.
Paul Rodman
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>
Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>
