Re: [boost] [SoC] Summer of Code Project Ideas

Subject: Re: [boost] [SoC] Summer of Code Project Ideas
From: Vladimir Prus (ghost_at_[hidden])
Date: 2010-03-09 16:00:44

Robert Ramey wrote:
> Vladimir Prus wrote:
>> Robert Ramey wrote:
>>> Vladimir Prus wrote:
>>>> Stefan Seefeld wrote:
>>>> Would you please list the tools that produce equally useful result
>>>> tables?  I am sure everybody will jump instantly as soon as such
>>>> tools are identified.
>>> I can't resist the opportunity to flog my own library_status program
>>> and library test scripts in the tools/regression/src directory.  I
>>> derived these from Beman's original compiler_status source.  I use
>>> it because it provides what I need in a fairly simple way.  That is
>>>
>>> a) it's built with C++ and boost so I didn't need to learn a whole
>>> new thing.
>>> b) it easily works with one library at a time on my local system
>>> c) produces a much more useful table with different columns
>>> for different build scenarios - release, dll, etc.
>>> d) works well with bjam
>>>
>>> It doesn't work with all libraries as it has to make certain
>>> assumptions about directory structure which are not
>>> true for some libraries.  It also has a fair amount of
>>> hacked code to try to accomodate these situations.  But
>>> all in all, I find it exactly what I need - No surprise there
>>> as I made it to solve my problems.
>> I am not sure what motifications you have made, but unless you
>> have rewritten everything from scratch I fail to see how
>> this can be a solution even remotely comparable to what we have
>> at present.
> 
> Hmmm - I'm not sure what you mean by "[not] comparable".  The points
> a-c describe how it's different - and  in fact superior - in my view.

The point if that we have reporting system having various features,
in particular display of result from multiple libraries, on multiple
toolchains, with expected failures, etc. You mention a solution
that you use for a single library, which works good for you. However,
that solution lacks most of features. Therefore, it cannot be used
to replace the current reporting system. And if it cannot replace
system, it's probably not important what programming language it
uses.

- Volodya