[Openmcl-devel] ...the sky is falling

Gary Byers gb at clozure.com
Tue Apr 6 09:53:16 UTC 2010



On Fri, 2 Apr 2010, Joakim Sandgren wrote:

> here is one console :sincerely
> joakim
>

Thanks.  I haven't been able to reproduce this, but can convince myself
that a bug that I did find could have led to this error.  Could you please
update to r13585 (in the trunk), rebuild the IDE, and let me know if this
problem persists ?

I think that the combination of factors that might lead to the bug is:

  - the definition you're looking for isn't at its recorded position in
    the file/buffer, either because the file/buffer has been edited since
    the function was defined or because that position is a byte offset (not
    a character offset) and the file is encoded in a variable-length encoding
    (like UTF-8).  (I don't remember if the latter possibility is a factor;
    Gail might know offhand.)  In this case, M-. has to search for the
    definition.

  - a data structure used to perform that search was initialized incorrectly
    if the length of the search string (the recorded source of the definition)
    was greater than the largest CHAR-CODE of the characters in that search
    string.  (In other words, it was completely bogus.)

That (mis-)initialization is definitely a bug, and it would have led to the
same symptom that showed up in the backtrace.

When reporting a bug, it's much, much better to send a backtrace (even if
it turns out to be unnecessary) than to not send one (if it does turn out
to be necessary.)  r13585 fixes -a- bug (and it's quite possible that it's
the one that's affected you), and it wouldn't have been possible to even
know where to look without the backtrace.





More information about the Openmcl-devel mailing list