[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: C++ transition stumbling blocks?

Junichi Uekawa <dancer@netfort.gr.jp> writes:

> > > API compatibility is not at issue. ABI compatibility is.
> > > 
> > Haven't you read the rest of my mail? I am _*not*_ talking about
> > binary package names here!
> Neither am I.
> 1. ABI incompatibile library packages should have different names
Different source package names? This means for if i have a source pkg
named 'foo' which produces binary pkg libfoo1, and upstream makes a
ABI change, I have to use a new source pkg name? This would be quite
against current practice, /methinks...

> 2. Packages which do not have corresponding source package will be 
>    removed from archive
Yes, but if there is no API change, dependant packages can be rebuilt
against the new library.

I think we are talking at cross purposes. Maybe the following can
clear it up:

Here is what Robert did:

id3lib (source)
  - libid3-3.7-13
  - libid3-dev

id3lib3.8.0 (source)
  - libid3-3.8.0-dev (provides, conflicts, replaces id3lib-dev)
  - libid3-3.8.0

id3lib3.8.2 (source)
  - libid3-3.8.2-dev (provides, conflicts, replaces id3lib-dev)
  - libid3-3.8.2

I think a better solution would be to have:

id3lib (source, as above)

id3lib3.8 (source, of upstream version 3.8.2)
  - libid3-3.8-dev
  - libid3-3.8.2

Packages depending on libid3-3.8.0 would simply be rebuilt against
libid3-3.8-dev (>= 3.8.2). It makes no sense (and wastes diskspace on
the mirrors + potentially diskspace *and* RAM on user's machines) to
have both libid3-3.8.0 and libid3-3.8.2 in the archive.

Regards, Andy
Andreas Rottmann         | Dru@ICQ        | 118634484@ICQ | a.rottmann@gmx.at
http://www.8ung.at/rotty | GnuPG Key: http://www.8ung.at/rotty/gpg.asc
Fingerprint              | DFB4 4EB4 78A4 5EEE 6219  F228 F92F CFC5 01FD 5B62

Reply to: