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

Re: gcrypt7/gnutls10 => gcrypt11/gnutls11 for Sarge

On 2004-07-24 "J.H.M. Dassen (Ray)" <fsmla@xinara.org> wrote:
> On Sat, Jul 24, 2004 at 17:35:59 +0200, Matthias Urlichs wrote:
> > * the API changes are minor and only require recompilation;

> "only [...] recompilation". Please provide details on the number of
> packages this affects. The last gnutls change affected 400+ packages
> (KDE, GNOME, cups, ...) and tooks months to sort out. More
> importantly, many end user applications were broken in the meantime
> as processes that happened to pull in both the old and the new
> version of gnutls crashed horribly. Is there any reason to assume
> this new transition will go faster and smoother?

Yes, there are two reason this should go smoother.

* The complete[1] gnutls7, gnutls10 (and the proposed gnutls11)
  dependency chains are using
  versioned symbols now.

* contrary to the last transition (libcupsys2 -> libcupsys-2gnutls10)
  different versions of gnutls and gcrypt can co-exist in the archive.

Of course the last transition was as bad as possible, so "smoother" is
not necessarily "smooth enough".
              cu andreas

[1] opencdk(|8), libtasn(1|1-2), libgcrypt(7|1). libgpg-error only
exsts in a single version and will contnue to stay this way, according
to upstream. I am not sure whether uploads of gnutls7/10 linking
against the new versioned dependencies are necessary.

"See, I told you they'd listen to Reason," [SPOILER] Svfurlr fnlf,
fuhggvat qbja gur juveyvat tha.
Neal Stephenson in "Snow Crash"

Reply to: