This is the mail archive of the mailing list for the libstdc++ project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

[Fwd: Re: sgi's std, stlport...]

and another:-)

-------- Original Message --------
Subject: Re: sgi's std, stlport...
Date: Wed, 5 Jul 2000 06:04:20 -0700 (PDT)
From: Dietmar Kuehl <>


after sending my answer on your request I noticed that I have not sent
it to the mailing list but only as a private answer to you. Thus, your
answer was also not sent to the mailing list...

> wouldn't it be useful to merge or somehow syncronize sgi's patch and
> stlport patch to stdc++ ? mainly in iostream and locale part.

SGI/STLport took a very different approach to implement IOStreams and
locales. ... and this is, in turn, again very different to the approach
I have taken for CXXRT. Although it is probably possible to share at
least parts of the implementations in this area, it is questionable
which part to choose and actually I think there is even and advantage
in having a variaty of different implementations.

> if they are really different and it's not possible to try to approach
> these implementations?

My personal thinking is that we should look at all three free
implementations at one point and join them. However, I don't think the
time to merge them has come already. In fact, I'm planning to diverge
even more than I have already done: I'm thinking of a new
implementation of the STL. Currently, there is basically only one
version of this library implemented: All STL implementations have the
HP STL at their root. I think it is quite reasonable to do certain
things rather different from the current implementation. ... and apart
from a possibly better approach this is actually also likely to reveal
weaknesses in the specification...

Variaty is not bad because it provides experience with different
approaches. I can definitely see advantages in my style to implement
the standard C++ library, eg. relatively short compile times. I think
it is reasonable to analyse the different implementations for their
respective advantages to create an even better implementation.



Do You Yahoo!?
Kick off your party with Yahoo! Invites.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]