Subject: Re: Potential 0.7.8 show stopper
From: Paul Rohr (paul@abisource.com)
Date: Sun Jan 30 2000 - 22:24:50 CST
Bob,
Thanks for doing all this testing. I'm glad to hear that the product is
holding up so well.
However, I have to disagree with you.
1. This behavior is not a crash, it's an assert. This difference is
important. Remember that all asserts get compiled out of release builds, so
users of our binaries will never see them.
In some cases, asserts will guard a subsequent crash. If so, that's
important to know. In other cases (such as this one) something's not being
done entirely properly under the hood, but as far as the user's concerned,
it all works out fine in the end.
2. This behavior *was* in 0.7.7. I know, because I left it there. ;-) If
you build a debug version of the 0.7.7 sources, you should see exactly the
same behavior.
3. This is definitely not a showstopper. We have had zero complaints about
this behavior in the nearly 2 months since 0.7.7 shipped. Yes, I should fix
the underlying problem (the fast squiggle logic doesn't do all the proper
spell checks when editing across block boundaries), but it's certainly not
worth holding the release for.
Paul
PS: I just tried a fresh clean build and couldn't even get the assert to
fire anymore. Now the only reminder that I have to implement that missing
code is that the last typed in a paragraph doesn't get spell checked.
Bummer. I miss that assert. It was a regular reminder that I should go
finish that code. Oh well. Now you've got a bug number to wave at me
instead. :-)
This archive was generated by hypermail 2b25 : Sun Jan 30 2000 - 22:19:29 CST