Boost logo

Boost Testing :

From: Victor A. Wagner Jr. (vawjr_at_[hidden])
Date: 2005-04-12 11:11:01

At Tuesday 2005-04-12 05:57, you wrote:
>Victor A. Wagner Jr. writes:
> > OK, nothing shows on the summary page, and I renamed the uploaded files to
> > to the uninspired names of save.........
> > hopefully you guys will find out why we're not finding the problems.
>Unfortunately (or perhaps fortunately?), the log looks consistent
>with the reports.
> >
> > unfortunately there aren't any errors in the log file...I'm trying to find
> > out what _I_ do differently in using boost::array than the tests (and you
> ^^^^^^^^^^^^
>Did you mean boost::range?

I didn't, because I don't directly call range.
I use... tuple, date-time, and array... perhaps I mistakenly judged it to
be from inside array.
<sarcasm>Microsoft in its _INFINITE_ wisdom, _innovated_ from the compiler
output, the points at which #included files are mentioned</sarcasm> which
means an error of the form
sfinae.hpp (25) some error message is almost worthless for finding the
CAUSE <extreme sarcasm>Thank you Microsoft</extreme sarcasm>

so I'm still looking for the problem.

In addition, I don't know what Jonathan Turkanis is doing with his editor
and / or cvs client, but I consistently get errors after he's updated things.
sometimes my client complains that it gets a checksum error after applying
a patch (only solution is to delete the file and let it do a U type
checkout instead of a P...which then causes the client to report an error
and the script will try 5 more times to get all the files (this adds around
20 minutes to the regression run time).
OR, I get line endings (on the lines he's changed) which
are \r\r\n (which my version of VC8.0 calls illegal)
A particularly gratuitous change is revision 1.5 of iostreams/chain.hpp
with the log entry:
revision 1.5
date: 2005/04/09 20:21:43; author: turkanis; state: Exp; lines: +5 -5
fixed whitespace

which ONLY added a trailing ' ' to ONE line and changed a bunch of line
endings to \r\r\n.
AND which triggered a remake for all the regression testers.

Getting this kinda crap to stop would certainly help.

> > can be sure I'll add a test when I do).
>Please do.
> >
> > there are many files which appear to use the range headers which aren't
> > compiling that far because of some missing ziplib.h or bziplib.h
> > files.... perhaps it was in that area that the problems existed the other
> > day.
>Just three of them, and again, all these failures are present in the
>Aleksey Gurtovoy
>MetaCommunications Engineering
>Boost-Testing mailing list

Victor A. Wagner Jr.
The five most dangerous words in the English language:
               "There oughta be a law"

Boost-testing list run by mbergal at