Am Dienstag 30 März 2010 20:59:58 schrieb Martin Paljak: > > * at least "pkcs11-tool --test --login" for already initialized cards. > > again some failures are ok (e.g. opensc trying some mechanisms on very > > restricted keys - for example signature cards restricted to few signing > > mechanisms, while opensc tries all mechanisms the card could support > > and doesn't know that this key is more restrictive). > > This is more like a smoke test. You get very bogus results and only "test" > a single slot. In the end, a real human needs to do a subjective test, if > a card (especially a pre-formatted card) works or not, in real life > applications (if it matters for that card)
well, knowing if there is at least a single person that could run a minimal test on a card, even that is much better than getting no feedback at all. and I think we should put a warning sign or at least some sort of hint on all cards, where we don't get any feedback. > Don't know if there are lots of resources for a systematic testing > procedures, especially if there are no automated tests that would actually > cover systematically more than a few specific use cases. for example web authentication with firefox isn't hard to test - "openssl s_server" with the right parameters is enough, so every unix/mac user should be able to test that, given a README, a supported and working card and some time to follow the steps. but I don't see how we can automate everything. if people can't spend half an hour to an hour to help us, there is little we can do. some support mails to the mailing lists to help someone take that much time to write.... > > i.e. if there are no active testers and developers for some plattform, > > we should people know, even if opensc is in the ports collection or > > something like that. for example debian doesn't upgrade opensc right > > now, because it doesn't compile on debian kfreebsd. but we haven't > > had any user from that plattform ever, and even for all *BSD I don't > > remember any active user. > > If you look at the real distribution of users in the Windows/Mac/Linux/*BSD > and combine it with the general availability or necessity of smart cards, > then this is the expected result. Cant beat the statistics. lets say, again we should have some place to let people know what works (was tested), doesn't work (got bug reports), or is unknown (no active users / feedback at all). > What should be done, is documenting exactly what we *know* that does not > work and vice versa, to the extent that makes sense. Until somebody > reports that it does not work on the Debian/kfreebsd, the fact that we > don't know if it works is not really relevant. for the record: openct on debian kfreebsd doesn't compile. and all the bugfixes etc. I send in for the debian package will not reach testing or stable, until someone can have a look (or openct is somehow excluded from kfreebsd, so it will be available at least on linux). Regards, Andreas _______________________________________________ opensc-devel mailing list opensc-devel@lists.opensc-project.org http://www.opensc-project.org/mailman/listinfo/opensc-devel
