On 16 November 2005 at 21:40, Steve Langasek wrote:
| severity 339255 important
| thanks
|
| On Wed, Nov 16, 2005 at 09:16:37PM -0600, Dirk Eddelbuettel wrote:
|
| > On 17 November 2005 at 03:22, Matthias Klose wrote:
| > | Compiler versions g++-4.0_4.0.2-4 and g++-3.4_3.4.4-10 are now in the
| > | archive. The renaming of the library packages can now start. You can
| > | upload the packages even before the toolchain is built for all
architectures
| > | because the packages with the new binary packages will be hold in the NEW
| > | queue until the required toolchain changes are installed on the buildd's.
|
| > I am getting this because of Quantlib. Now, for libquantlib-0.3.11, the most
| > recent version, as well as for the preceding ones:
|
| > [EMAIL PROTECTED]:~> apt-cache rdepends libquantlib-0.3.11
| > libquantlib-0.3.11
| > Reverse Depends:
| > r-cran-rquantlib
| > quantlib-ruby
| > quantlib-python
| > quantlib-examples
| > quantlib-examples
| > libquantlib0-dev
| > libquantlib0-dev
|
| > Do I really need to do the "c2a" renaming dance? I upload a new Quantlib,
| > and a day later rebuild my two packages that depend on ? [ quantlib-swig
| > provides quantlib-ruby and quantlib-python; rquantlib provides
| > r-cran-rquantlib; quantlib-examples comes fromq quantlib itself. ]
|
| Do you *need* to? Technically, since this library soname has been confined
| to unstable and presumably has not been used elsewhere as a .deb under this
| name, it's not an RC bug if the ABI changes without a change in package
| name. It's still a bug, of a sort I believe we should take seriously; but
| it won't prevent quantlib from shipping with etch...
|
| > This small set makes Quantlib a little easier to deal with than a full blown
| > KDE component or base library. So shall we do this without "c2a" ?
|
| No, the size of the reverse-dependency tree is not a factor in whether the
| bug should be treated as release-critical. If a package is presenting a
| shared library interface, this must be done responsibly so that *however*
| many packages there are depending on it, they don't break sa a result of
| partial upgrades.
I will interpret "done responsibly" as tacitly supporting my plan suggested
above and upgrade in a two-step
day 1 quantlib
day 2 quantlib-swig, rquantlib
Unless of course I hear screams of bloody murder from either of you before
this eve (Central time, that is). :)
Thanks as always for all your work on Debian (and Ubuntu, in doko's case).
Regards, Dirk
--
Statistics: The (futile) attempt to offer certainty about uncertainty.
-- Roger Koenker, 'Dictionary of Received Ideas of Statistics'
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
