ext/ names (was: re:[PATCH] HP/SGI extensions to...)

Nathan Myers ncm-nospam@cantrip.org
Thu Jan 3 11:27:00 GMT 2002

> >> this is the next round. In fact we are almost done, only a few bits
> >> belonging to stl_tempbuf.h and stl_tree.h still are to be dealt
> >> with. Do you have any particula suggestion for the most suited ext/
> >> name? Perhaps ext/misc ?
> >
> >maybe ext/tree?

Completely aside from the question of whether we should be including 
<algorithm> or <bits/std_algorithm.h> in our sources -- to be resolved,
I hope, according to engineering consequences -- I object to naming new 
headers without a meaningful extension.

The standard committee hit on leaving off the extension as a trick 
to sidestep the politically inflammatory issue of whether its headers 
should have a ".h", ".hpp", ".hxx", ".hh", or other suffix.  Their 
problems are not our problems.  Their weaselly example is not a good 
example for us.  We have no need for weaseling.  

If we define local headers, we should give them proper suffixes, so that 
the set of files without is constrained to the well-known set specified 
in the standard.  I see no reason for any name under ext/ or other local 
directories to lack an identifying suffix. 

Besides the inconvenience suffixless names cause us, it also sets a bad 
example for users.

Nathan Myers

More information about the Libstdc++ mailing list