Craig A. Berry wrote:
I will look at these, test them, and -- if all goes well -- apply them,
but it's likely to take me a while to get to them. In general I agree
with the principle that routines that can be called very early in
start-up before thread context is initialized do need to be cautious
about trying to access that context. I will just make a preliminary
comment that it's very hard to see what's going on by reading the
patches because they consist mostly of changes to non-significant
whitespace.
Thanks for looking at these.
It may be easier to apply the patches and then compare the results
against the unpatched vms.c manually.
The diff program tried to find common parts and did not synchronized well.
I will not have the last big patch ready until tomorrow. I found some
more cases with the null context, and more that did not need a context
at all. So I need to run through a test cycle.
The upcoming patch also has the some fixes for Unix filename handling
that I missed breaking out separately. Primarily cases where directory
specifications returned by glob and other places need to have the ".dir"
removed.
It will also has a fix for trim_unixpath, which has a memory leak.
-John
[EMAIL PROTECTED]
Personal Opinion Only
