Hello, On Aug 30, 2010, at 10:22 AM, Jean-Michel Pouré - GOOZE wrote: > IMHO, I support a 0.12.0 release for these reasons: > > * Maintaining two different releases is a lot of work. ... Yes, it is apparently difficult and a lot of work to roll out a 0.11.14 tarball (with the pcsc-lite changes), as it has not yet happened ;) But that would be a final source drop for 0.11. X. It is not maintaining them separately as Linux 2.4 and 2.6 co-existed (still do?) for a looong time [1].
> * Maintaining two different releases slows down the adoption of new > releases. Take the example of OpenSSL, which maintains 0.98 and 1.0 OpenSSL example does not apply, as those are direct linking dependancies. With direct linking it could go on forever, as early versions of OpenSSH for example allowed linking against libopensc (up to 0.11.X), which is from 0.12.X onwards not supported, and from OpenSSH 5.3 onwards not needed. And seriously - if OpenSSL 0.9.8 works for you, why bother upgrading? I doubt essential that stuff like verifying RSA signatures in small volumes or calculating SHA256 has changed a lot between 0.9.8 and 1.0.0. > Usually, the latest release never makes it in GNU/Linux distributions. > So you get stuck with an old bug fix release and users with binaries > versions rarely enjoy recent versions. A 0.11.14 would be exactly for such cases - people who are stuck (for example, because they link OpenSSH against libopensc) but for some reason need critical/essential bugfixes and updates. > * Free software is fast evolving. Releasing OpenSC 0.12 deprecates > 0.11.x and in a way will force package maintainers to publish a new > version of OpenSC in distros. Same for dependent projects. See below about forcing. > * 99% of GNU/Linux users use binaries, and rarely compile from sources. > Except people on the mailing list! So releasing 0.11 bugfix would mean > that a lot of people would not use 0.12 and therefore would post fewer > bug reports. Providing "official unofficial" .deb and .rpm packages would be nice (as said in a previous e-mail). Feel free to work on that. As said, 0.12.0 changes internal package structure (at least on Debian based distros) which require extra work from package maintainers. Feel free to provide reference packaging scripts for 0.12, so that the adoption by Debian or Ubuntu would be easier. Nothing can force a distro maintainer to upgrade packages to the latest upstream version. And nothing prevents a distro maintainer from packaging pre-releases or SVN versions. But I'm sure distro maintainers are reasonable people and understand the need to upgrade to newer versions, whatever their version number, if correctly requested. And that will happen faster if some work has been done for them beforehand (like packaging scripts they can re-use and/or improve) > * The current situation is that a lot of "supported smartcards" are > either end-of-live or not supported any longer or you can simply not buy > them as they are not produced. On the converse eID is on the rise. Correct. > Releasing 0.12 and then incorporating Viktor's changes will clarify the > situation. We need fewer smartcards, but with excellent support. And > frequent OpenSC releases. Almost correct (we as many cards as possible, but all with good or excellent support) > For all these reasons, I am against a 0.11.x bugfix release. Sorry, I don't see how this could conflict with 0.12.0 release/adoption. [1] http://amailbox.org/Linux/Who_Uses_The_2.4_Stable_Kernel -- Martin Paljak @martinpaljak.net +3725156495 _______________________________________________ opensc-devel mailing list opensc-devel@lists.opensc-project.org http://www.opensc-project.org/mailman/listinfo/opensc-devel
