[Boost-users] Solving boost version issues

From: Daniel Krügler (dsp_at_[hidden])
Date: 2006-11-02 07:32:32

Hello,

I apologize in advance, if this contribution is slightly off-topic, but
I assume that the discussed problem is of general interest for most
boost users.

Not rarely one stumbles across the problem that even your most favourite
3rd party library has an implementation error, and you need a fix for
that. And the good thing: Those highly motivated guys have already found
the error and fixed it - in cvs.
Well, that is actually all very fine, especially if you are able to 
handle with cvs (who of the programmers can not?) and you could use this
fixed version, but there remains one big problem: The 3rd party library
has not been released and you want to prevent nasty effects, if your
fixed version meets the older version - the have all the same lib names.

OK, end of the scenario, I assume that I describe here a well-known and
well-feared problem ;-)
Actually everything is fine, besides...: I would like to modify the lib 
names and which place would be more reasonable than the bjam configuration?

First, because the described problem just hits me in the moment 
concerning a boost fix and I would like to influence the current bjam, 
such that all libraries are build as usual, but with an additional 
user-defined (or even hard-coded(!) string constant (aka "build id") at 
the end of each name. I have to confess that I have little experience 
with bjam and I don't understand the way how I can modify the related 
file(s) to reach the wanted effect: I can only assume that one has to 
change the file Jamfile.v2, probably the "function" rule tag beginning
with line 95.

Please, is someone willing to help me in that situation?

Second I would like to propose that the current bjam script is generally
modified to allow the addition of such a user-defined build-id simply
attached after the normal naming scheme described on

http://boost.org/more/getting_started.html#Results

libboost_date_time-gcc-mt-d-1_31-your_build_id.a

with a new, to be defined bejam option, which allows such an addition.

Sorry for the noise and thank you for your patience!

With Greetings from Bremen,

Daniel Krügler