Jim Meyering <jim <at> meyering.net> writes:

> > http://repo.or.cz/w/gnulib/ericb.git?a=shortlog;h=refs/heads/hash
> > $ git pull git://repo.or.cz/gnulib/ericb.git hash
> 
> By the way, in this change set,
> 
>   http://repo.or.cz/w/gnulib/ericb.git?a=commitdiff;h=bf740889e7ca68be9e
>   fake memory failure, for testing hash_rehash

Well, I _did_ mention that the commit was for testing purposes only, and not to 
be committed to savannah ;)

> 
> I noticed that this is a no-op:
> 
> +            if (i == 363)
> +               i = i * (i - 362);

Not quite.  It exists to provide me a breakpoint that won't get optimized out, 
and will still trigger prior to starting the offending hash_rehash, as 
determined by the previous run where abort() triggered after hash_rehash had 
already run out of memory.  But yes, it makes for an awkward-looking gdb 
breadcrumb.

-- 
Eric Blake





Reply via email to