Build, Release, QA
------------------
*Change to Linux full build*
Bear explained that the linux full build no longer uses OSAF's openssl,
the libssl-dev package is now required.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008481.html
*Readme*
Sheila gave the text for the readme files, explaining that the file
gives a link to the launch page and little other information. She asked
if the readme should contain any other details.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008480.html
Philippe argued we should continue to keep it short with links, avoiding
duplicating information.
- how to install and run for the first time
- reminder to look at relnote/open issues
- pointer to the license file
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008499.html
*Functional tests, -O*
Grant observed that rt was running functional tests with -O. The script
recording framework currently uses asserts for verification, so no
verification is being done. Also, non-test code that triggers asserts
will be ignored instead of causing failures.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008482.html
Grant pointed out an earlier decision where debug tests are always to be
run without -O, and release tests only to be run with -O when run with
debug tests. Heikki noted that the fix didn't get ported to rt, Bear
logged a regression.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008488.html
Grant, Heikki and John debated whether or not tests should rely on
assert. Heikki and Grant dislike using asserts as part of the test
framework, as one wants to be able to invoke tests with -O and not break
the framework. John likes using assert because it is convenient for
debugging, and is a standard mechanism. The discussion gets going about
here in the thread:
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008491.html
Heikki proposed raising AssertionErrors if test conditions fail, guarded
by a flag. Grant suggested wrapping everything in a function.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008493.html
Mikeal elaborated on the debugging issue: the debugger breaks where the
assert gets called, not in a test library. Mikeal agrees we need to move
away from assert in the test framework, and is open to suggestions that
satisfy the debugging requirement.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008504.html
*Localization string code freeze*
Philippe scanned the bugs, listing the ones that will affect the
localization string freeze. Any strings (requiring translation) that
need to be added or changed should be handled before the 9th.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008508.html
*Wiki*
Katie proposed changes to desktop related wiki pages, marking some as
obsolete and moving some from being "project pages" to being
documentation or release oriented planning.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008507.html
Heikki objected to marking the security page as obsolete.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008509.html
*Test failures*
John and Dan tracked down the RenameAndNewEvent test failures to be
related to size changes affecting the minical and sidebar. Philippe
asked if the test framework could be more robust in handling platform
specific size differences and mouse coordinates.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008514.html
John replied that mouse coordinates are rarely relied upon -- this
particular failure was due to the collection being partly obscured.
Philippe explained that the divider capture is based on mouse coordinates.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008516.html
Bear changed the initial window size for perf tests, fixing failures.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008518.html
Dev
---
*timing utility*
Morgen proposed removing util.timing, noting that it is used by a
handful of tests. Heikki replied that the numbers gathered are not used,
so its ok to remove all uses of util.timing.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008498.html
*OPTIONS not PROPFIND*
Morgen announced that Chandler now uses OPTIONS instead of PROPFIND when
starting a subscribe operation. Post-preview he'll look into removing
the HEAD request.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008501.html
Meetings, Announcements
-----------------------
Bug council:
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008502.html
IRC QA session (Cosmo 0.6.1.1):
http://lists.osafoundation.org/pipermail/chandler-dev/2007-July/008512.html
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev
