From: Jeff Garland (jeff_at_[hidden])
Date: 2003-10-06 08:36:25
On Mon, 06 Oct 2003 09:35:06 +0200, Jonathan de Halleux wrote
> Thank you for your positive answers about the zipped streams. Here's
> a summary of the suggestions I received:
> 1 - yes, zipped streams could be useful in boost,
> 2 - "is it appropriate for a boost library to require another
> (excluding the standard library) to use" (Jeff Garland). ¨ Personaly,
> I think it is not a problem. zlib or bzip2 are already used in a
> looooot of applications with problem. Of course, boost people should
> give their oppinion on this,
I tracked down the discussion in the archive from Feb 2001. I believe it led
to some changes in the policies page on the subject of library reuse. Since
then a whole page has been written.
Conclusion it should be ok.
> 3 - A third suggestion was asking for user defined compression
> (Ehsan Akhgari, Pavel Vozenilek ): basically, the wrapper is not
> designed for that right now. There are a few reasons for that, that
> could be discussed further: The zlib and bzip2 libraries have
> some intrisic differences: their constructor arguments, the number
> of configuration parameters, zlib can add a header, bzip2 not,
> etc... All these differences would make a common wrapper quite
> difficult to make. Anyway, all suggestions are welcome.
In my wrapper stream class I adapt and expose some of the underlying API
options -- you could do something similar.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk