Re: [boost] [RFC] Better error messages for errors detected atcom pile time

From: Daniel Frey (daniel.frey_at_[hidden])
Date: 2002-03-26 05:16:22

Douglas Gregor wrote:
> 
> On Monday 25 March 2002 07:39 pm, you wrote:
> > I've heard of, but never used Leor Zolman's STL Error Message Decryptor
> >
> > <http://www.bdsoft.com/tools/stlfilt.html>
> >
> > uses perl.

Maybe you want to have a look at booster.zip in the files section. It
got almost no attention from this group so far but it may be the time
for a second chance :) An updated version is available if anyone would
like to see it. I'm using it for a number of things here in the company:

a) Convert long STL-names to readable names, e.g. std::map< std::string,
std::string >
b) Convert (some) long boost-names to readable names, e.g. boost::regex
c) Remove warnings from external libraries about comparing int's to
unsigned int's, etc. that are useless but always there. Roguewave (as an
example) has warnings about reordering the initialization list of some
ctor's but they refused to change the code). Instead of patching there
code, I just added a regex to remove the warning from the compiler
output.
d) Replace our internal typedef-ed names (RWDecimal<
RWMultiPrecisionInt< 3 > > becomes Base::decimal, etc.)
d) Remove warnings from the make-system and colorize the output to
better see what's happening
e) Summarize and colorize CVS-output to highlight conflicts during
updates, etc.

> Simple summary: STLfilt works right now, and doesn't change the language but
> what I'm proposing can help make much nicer error messages.

The problem is, that you can't change the compiler/language. But you can
write a macro to output strings from the code as a warning from the
compiler. These warnings can be used to generate good error messages.
Sure the tool I wrote is not yet ready for these kind of things, there
are several extension that could be useful: Regular expressions that are
applied to several lines at once (allowing to summarize the call stack
and spread errors - that is errors that can only be detected when
looking at several lines of output at once). Maybe includes for the
rulesets would be useful. You may have a look for yourself and see if
you like it...

Regards, Daniel

--
Daniel Frey
aixigo AG - financial training, research and technology
Schloß-Rahe-Straße 15, 52072 Aachen, Germany
fon: [phone], fax: [phone]
eMail: daniel.frey_at_[hidden], web: http://www.aixigo.de