[Date Prev][Date Next][Subject Prev][Subject Next][ Date Index][ Subject Index]

RE: Memory Leak?



** Reply to note from xywrite@xxxxxxxx Wed, 7 Apr 1999 09:36:56 -0500

> Then why did you complain about what I was doing before as being an
> "illegal use of wildcards"?

Because, due to capital "W" and capital "R", my initial thought was that you had
resorted to using a wild-X to represent an Ascii-10 (your "LF"), because you
couldn't get a real Ascii-10 onto the CMline (originally, if I'm not mistaken, you
called it a "character"; but there is no Ascii-10 character that can be put on the
CMline -- in Xy3, yes; but in Xy4, no). In fact, the more I thought about your
note, the less I was sure what you meant, in part because you didn't describe it
accurately. Hardly a "complaint" on my part -- merely a suggestion.

Now, since you're having the trouble ("complaining") and I'm not, and we're both
executing the same command (apparently), then it seems to me that it must either
have something to do with the resulting content of your page (after replacements),
or, possibly, with a memory buffer being right on the edge of being filled before
the command is issued, and over the top after it is issued. As an example of the
first, I can construct a search/replace command that will fill my CMline and screen
with STACK STACK STACK STACK and force me to shut XyWrite down. As for the second,
a large PRiNter file, with lots of SUb and WiDth tables, can force all kinds of
strange things to happen -- the misbehavior of commands having nothing to do with
PRN. You need to test lots of things. For example, load STRIP.PRN instead of
your usual PRN, and see if it makes a difference. When you command "VA/NV
$M+6", what does the PRompt line say (both before and after issuing the
command)? What does VA/NV $ER say after the command crashes? Check your memory
page in DLG and see if any of the reports look high (put "JM 0mQ2 " in a dummy
file, DeFine it, then RUNCODE it).

>> My experience is, that if you obey the rules, XyWrite obeys the rules; but
>> that Xy4 can get extremely stressed when you push its parsing capabilities
>> to the edge.

> That's nice in theory. But I don't see why what I was doing is "outside
> of the rules" of XyWrite.

I'd wager that you ARE doing something wrong. That you don't see why it's wrong
(or what it is) is self-evident. But eventually you'd probably find
something, if you hammered at it. I mean, if you look back over the problems
that are discussed here, people get real mad, defensive, blame the program, and
leap to self-exonerating conclusions; but 99% of the time what you find out in
the end is that they screwed something up. So it's a reasonable
presumption -- don't you think?

What's the content of the {note}? What does {qst} mean? Just for the
sake of argument, suppose there's a three-byte extended Speedo character 256-909 in
there somewhere, or anyway something with an Ascii-254 in it. Put the right
sequence of characters inadvertently together, and you can trigger all sorts of
wierd screen behavior (because 254 is the main screen
handler) and crash your session.


-----------
Robert Holmgren
holmgren@xxxxxxxx
-----------