lsbsi Jan 7, 2003 Confcall Notes
"Wichmann, Mats D" <[email protected]>
| Newsgroups | gmane.linux.lsb.implementation |
|---|---|
| Message-ID | <[email protected]> |
NOTE: effective Jan 14, the confcall will occur one hour earlier (11am Mountain time, immediately after the lsb-futures call). Please update your calendars accordingly. Beta lsbsi bits have been uploaded to the freestandards.org impl/beta/binary directory. These are tagged 1.3b1, and are available for ia32, ia64 and ppc32. The "test" packages, allowing the test suite to be used, still need to be uploaded. The impl/beta/source directory will be updated soon. After a review that the webpage adequately describes what to do with the lsbsi and how to use it, from the viewpoint of someone who's not involved in this project, the availability of this beta test will be announced soon. There are remaining i18n failures. These will be documented and left, for now; they may be addressed in an update release later if there's movement with the upstream sources, or if someone suddenly finds time to work on the patches, many of which are available from the team, but against different versions of the upstream releases. Testing is currently in progress with the latest beta runtime test suites, results will become part of the lsbsi release notes. Stuart and Mats are both running these tests. The other main test that has to occur is a cross-check that all the 1.3 appbat packages install and run, for each architecture. Marvin and Mats will work on this. ia32 is already built on multiple distributions and ia64 on at least two, so there's some confidence of portability. One additional ppc32 distribution will be tested soon as a sanity check. A strange new problem has cropped up with the nALFS tool, it seems that in some circumstances, if logging is turned on (perhaps affected by one other flag), the child nALFS process dies. Since avoiding "-l -v" seems to avoid the problem this is low-priority at the moment. Someone should file a bug (hint, hint). There was also some discussion of a few minor tweaks to make the "lsbsi-test" environment easier (mainly, making sure an /etc/services file is present and a few other setup tweaks); these will be added for beta 2 (if there is one) or the final - they are low-impact. Nicer would be to have the lsbsi-test pretend to "boot", as in starting init and kicking off a login process. This will be invesitated. We'd also still like to get the UML lsbsi going. This is back burner because it's not needed for any testing/certification program, but would make for a much nicer environment and package.