[Bug libstdc++/45841] [4.6 Regression]: r164529 cris-elf libstdc++ 27_io/basic_filebuf/seekoff/char/2-io.cc

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45841

--- Comment #38 from Paolo Carlini <paolo.carlini at oracle dot com> 2010-10-05 23:34:37 UTC ---
> To summarize the comments above, the real issues I know of at r164529 are:
> 1) an extra lseek (compared to r164529) for
> 27_io/basic_filebuf/seekoff/char/2-io.cc
> 2) erroneous behavior that David found, when reading past the end-of-file

Ok, thanks for the summary. Thus it looks like we are in good shape (David said
in a previous message that 1) isn't really worth optimizing for, if I remember
correctly).

> (In reply to comment #34)
> > Posted a patch to fix my end of this, and a regression to verify that fix on
> > working systems.
> > 
> > http://gcc.gnu.org/ml/libstdc++/2010-10/msg00015.html

Good, I'm going to apply it here, give it a spin and commit it if everything
looks fine.

> <nitpick>
> David, regarding contracting the expression "regression test" into
> "regression": Don't.  It changes the meaning in a bad way: you add a
> "regression *test*" not a "regression".  I hope; at least eventually. :)

I must say I totally agree. More substantively, please, at your ease, figure
out a more complete and independent testcase. The normal rule is: one bug
report, one testcase.