From Apache OpenOffice Wiki
Revision as of 07:27, 23 July 2007 by Jsi (Talk | contribs)

Jump to: navigation, search

These are the IRC minutes of the second Automation meeting on IRC found place at Monday 16th July

        • BEGINNE LOGBUCH UM Mon Jul 16 08:52:25 2007

Remove branch tags from all non-working and unused files for the upcoming 2.3 branch. (JSK)

Jul 16 09:00:22 sunrayjogi	Welcome to the second Automated GUI Testing IRC meeting!
Jul 16 09:00:53 sunrayjogi	The Agenda is taken from the WIKI (see IRC title) and we start with skotti's point
Jul 16 09:00:54 sunrayjogi	1. Remove branch tags from all non-working and unused files for the upcoming 2.3 branch. (JSK)
Jul 16 09:00:59 sunrayjogi	skotti:  your turn
Jul 16 09:02:08 skotti	sorry for the delay
Jul 16 09:02:43 ExLachs_home	usually you just delete the file in CVS so you can still check out older versions of the branch
Jul 16 09:02:59 skotti	I wanted to suggest that we remove all 2.3 tags from the tests that do not run on a 2.3.
Jul 16 09:03:09 sunrayjogi	ExLachs_:  please wait until skotti has described his issue
Jul 16 09:03:45 sunrayjogi	All: Any comments?
Jul 16 09:03:46 skotti	This has been done by sun_msc and me last time and made it easier for the guys outside sun to distinguish working and non-working tests
Jul 16 09:03:57 maho	great
Jul 16 09:04:31 skotti	So if somebody checks out the branch, all tests from that set of files are expected to work
Jul 16 09:04:40 sunrayjogi	All: Additional comments?
Jul 16 09:04:44 skotti	What do you think?
Jul 16 09:04:51 sunrayjogi	skotti:  Thx, I think we have all understood what you meant.
Jul 16 09:05:11 maho	has 2.3 branch  been created?
Jul 16 09:05:12 ExLachs_home	great idea expect that the information that the test once was part of the branch gets lost
Jul 16 09:05:32 skotti	maho: No
Jul 16 09:05:45 sunrayjogi	If a test is not usable in 2.3 why not removing it from the branch?
Jul 16 09:05:46 skotti	ExLachs_home: Yes, but does it matter?
Jul 16 09:05:47 maho	okay, then where do you annouce? dev@qa?
Jul 16 09:06:12 sunrayjogi	maho:  why announcing and what?
Jul 16 09:06:26 ExLachs_home	I don't know, but if the branch has not been created enyway there is no informatuion loss of course
Jul 16 09:07:10 ExLachs_home	its always worth a thought before you throw away information if there is not another way to preserve it and get the same results
Jul 16 09:07:13 maho	for example, ooo230 is tagged in testscripts, then users can test OOo 2.3 with it
Jul 16 09:07:35 maho	sunrayjogi : for example, ooo230 is tagged in testscripts, then users can test OOo 2.3 with it
Jul 16 09:08:06 skotti	ExLachs_home: The branch will probably be created at some point. But i don't see any problem in junking the branchtags on non-working files.
Jul 16 09:08:11 sunrayjogi	maho:  don't understand. If I have tests which won't work on a revision/release, I delete them
Jul 16 09:08:32 sunrayjogi	skotti:  You have "non-working files" now in HEAD?
Jul 16 09:08:49 skotti	ExLachs_home: We have a list of working tests out, but the list is held current, when somebody wants to test a 2.3 at a later time he won't find the information anymore.
Jul 16 09:08:51 maho	sunrayjogi : for 2.3, we use testtool on HEAD?
Jul 16 09:09:01 sunrayjogi	maho:  no
Jul 16 09:09:11 maho	sunrayjogi : then, which ?
Jul 16 09:09:16 skotti	sunrayjogi: No?!?
Jul 16 09:09:47 sunrayjogi	maho:  we will create a branch if the first CWSes arriving at QA which differs from HEAD and OOG (?)
Jul 16 09:10:18 sunrayjogi	maho:  that have we done all the time. The development is going on, so we need a branch for the test cases, too
Jul 16 09:10:29 maho	sunrayjogi : okay, if you create a branch please announce at somewhere. my recommendation is at dev@qa.
Jul 16 09:10:39 maho	sunrayjogi: okay
Jul 16 09:10:41 sunrayjogi	maho:  "testtool" means in this case "qa/qatesttool" in CVS, am I right?
Jul 16 09:10:48 maho	sunrayjogi: right
Jul 16 09:10:55 sunrayjogi	maho:  yes, we have done that all the time
Jul 16 09:11:01 maho	sunrayjogi: thanks!!
Jul 16 09:11:15 sunrayjogi	All: Back to the suggestion from skotti - Any comments?
Jul 16 09:11:33 sunrayjogi	skotti:  You have one question not answered
Jul 16 09:11:48 maho	+1 for removing scripts that are not used
Jul 16 09:11:49 sunrayjogi	skotti:  If a test is not usable in 2.3 why not removing it from the branch?
Jul 16 09:12:09 sunrayjogi	maho:  we will do that after we have closed discussion, thx
Jul 16 09:12:18 maho	okay :)
Jul 16 09:12:52 ExLachs_home	sunrayjogi, what exactly "is remove it from tha branch" you can eather remove the branchtrag or mark it as dead on the branch
Jul 16 09:12:54 skotti	sunrayjogi: I guess it's answered at this point.
Jul 16 09:13:12 skotti	ExLachs_home: Remove the branch tag
Jul 16 09:14:08 sunrayjogi	ExLachs_home:  We are not talking about code for software. We are talking about code for tests which matches to ONE release of a software under test.
Jul 16 09:14:43 sunrayjogi	ExLachs_home:  skotti , as far as I understood him, wants to remove the branch tag from a .bas/.inc/... if the test does not work in that branch
Jul 16 09:15:03 ExLachs_home	sunrayjogi, ah, ok so you really don't need the prerelease information there
Jul 16 09:15:09 sunrayjogi	ExLachs_home:  I suggested to DELETE the files FROM THE BRANCH because then you won't have it on your disk (the same result)
Jul 16 09:15:24 sunrayjogi	ExLachs_home:  Yes
Jul 16 09:15:33 skotti	sunrayjogi: It would actually be sufficient to remove the branchtag from the .bas-file. But doing so also for .inc-files is cleaner
Jul 16 09:15:44 sunrayjogi	All: Other understandings of what I have summarized?#
Jul 16 09:15:48 ExLachs_home	sunrayjogi, deleting was what I suggested too
Jul 16 09:16:02 sunrayjogi	ExLachs_home:  ok
Jul 16 09:16:31 sunrayjogi	skotti:  IMHO you should not have unused inc-files NOW in the HEAD as well as in BRANCH...
Jul 16 09:16:50 skotti	sunrayjogi: It doesn't matter for the topic at hand.
Jul 16 09:17:00 sunrayjogi	skotti:  Removing the tag can be forgotten, deleting a file not. Be sure :-) I prefer deleting...
Jul 16 09:17:49 sunrayjogi	All: Okay, no other sugestions.... ? I will summarize what has been written as I understood
Jul 16 09:18:01 ExLachs_home	deleting has the advantage that a file is even deleted if someone does a cvs update
Jul 16 09:18:19 ExLachs_home	if the tag is removed the file will just remain there
Jul 16 09:19:27 sunrayjogi	1. Agenda issue: skotti has suggested to remove the branch tag from tests/files if they are not matching for that release (in that branch) e.g. ooo230 branch will a set of test scripts for OOo 2.3 and if a test / file in qa/qatesttool is not valid for that release remove the branch tag.
Jul 16 09:20:12 sunrayjogi	2. sunrayjogi suggested to delete the file in the branch. The file won't be checked out, even if CVS UPDATE is being made (ExLachs_ * said)
Jul 16 09:20:46 sunrayjogi	All: Please vote for 1. or 2. - first we vote for 1. with "+1" (pro) or "-1" (contra)
Jul 16 09:21:14 maho	-1 for 1
Jul 16 09:21:24 sunrayjogi	-1 for 1
Jul 16 09:21:25 oliverc	-1 for 1
Jul 16 09:21:28 b_osi	-1
Jul 16 09:21:59 ExLachs_home	-1
Jul 16 09:22:02 skotti	sunrayjogi: Removing the branch-tag corresponds to deleting the file from the branch, there is IMHO no difference between the two suggestions, however: -1
Jul 16 09:22:19 *	sunrayjogi 3....
Jul 16 09:22:23 *	sunrayjogi 2.....
Jul 16 09:22:30 *	sunrayjogi 1.....
Jul 16 09:22:33 sunrayjogi	okay
Jul 16 09:22:48 sunrayjogi	Voting for 2. : +1 for pro; -1 for contra
Jul 16 09:22:54 ExLachs_home	+1
Jul 16 09:22:56 b_osi	+1
Jul 16 09:22:58 sunrayjogi	+1
Jul 16 09:23:01 maho	+1
Jul 16 09:23:01 skotti	+1
Jul 16 09:23:06 oliverc	+1
Jul 16 09:23:15 *	sunrayjogi 3......
Jul 16 09:23:22 *	sunrayjogi 2......
Jul 16 09:23:27 *	sunrayjogi 1......
Jul 16 09:23:31 sunrayjogi	DONG :-)
Jul 16 09:24:20 sunrayjogi	Okay, I will announce it that files in qa/qatesttool which are not valid for a release have to be removed otherwise an issue can be written to the owner of the file/test. next agenda issue

Get rid of "The first doc!" window? (JSI)

Jul 16 09:24:32 sunrayjogi	2. Get rid of "The first doc!" window? (JSI)
Jul 16 09:24:48 sunrayjogi	Situation today... we open a document and modify it.
Jul 16 09:24:57 sunrayjogi	This document is always present
Jul 16 09:25:33 sunrayjogi	I have had problems under X11 (Solaris SPARC 9) that instead of a loaded document in the test this "first doc" has been used for export.
Jul 16 09:25:38 sunrayjogi	The test failed.
Jul 16 09:26:05 sunrayjogi	There is (AFAIK) no parameter for X11 available to guaranteed get the focus into a window.
Jul 16 09:26:39 sunrayjogi	It would avoid problems if we stop the "first doc"-routine in HEAD (not for 2.3!) to get it done until 2.4.
Jul 16 09:26:47 *	sunrayjogi as moderator
Jul 16 09:26:52 sunrayjogi	All: Comments
Jul 16 09:27:30 ExLachs_home	the alternative would then be to have a backing window only at teststart?
Jul 16 09:27:32 skotti	sunrayjogi: This has another advantage that we would finally be able to test the office state without the options settings for Writer
Jul 16 09:27:42 sunrayjogi	ExLachs_home:  yes
Jul 16 09:28:14 oliverc	Maybe someone should explain why this "feature" (first doc) was implemented
Jul 16 09:28:20 sunrayjogi	sunrayjogi:  mhhh... that is also possible today if you close that window in your test....
Jul 16 09:28:29 skotti	sunrayjogi: The disadvantage is that (especially in framework) the check for the first doc does not work anymore. That's some 30+ Tests to change
Jul 16 09:28:30 sunrayjogi	oliverc:  yes.
Jul 16 09:28:47 sunrayjogi	skotti:  That was reason to say 2.4
Jul 16 09:28:56 sunrayjogi	skotti:  We all have to change tests.
Jul 16 09:29:06 skotti	sunrayjogi: I know. I just wanted to inform about the impact.
Jul 16 09:30:20 skotti	sunrayjogi: There is an alternative though - that would be to enhance hNewDocument to allow actually closing the first doc thus treating the backingwindow just as any other documenttype
Jul 16 09:30:22 sunrayjogi	oliverc:  The reason was (if I remember right) that we would otherwise close the office completely without knowledge (sometimes). In the past we did not have the DocumentCount() routine in TT.
Jul 16 09:31:03 sunrayjogi	skotti:  From my point of view we have only to delete thefirstdoc() routine and have to check our tests. No other changes needed.
Jul 16 09:31:12 sunrayjogi	skotti:  no other GLOBAL changes
Jul 16 09:31:24 sunrayjogi	skotti:  the test maintainers have to do some work...
Jul 16 09:31:48 sunrayjogi	All: Other comments?
Jul 16 09:32:07 skotti	sunrayjogi: Quite possible. I just wanted to offer an alternative suolution in case someone turns up with a reason why your suggestion might be troublesome
Jul 16 09:33:04 sunrayjogi	skotti:  Could you please explain?
Jul 16 09:34:26 oliverc	So, the main reason why the first doc was introduced has vanished and we have focus problems with the first doc => get rid of this routine
Jul 16 09:34:41 skotti	sunrayjogi: I rember running into a scenario where i tried to work with the backing window and ran into trouble - just can't remember what it was. 
Jul 16 09:35:19 sunrayjogi	oliverc:  no, we didn't want to close the whole application :-)
Jul 16 09:35:52 sunrayjogi	skotti:  That would be an issue but doesn't hinder us to decide removing the additional (faking) routine
Jul 16 09:35:53 skotti	oliverc: There alwas was the reason for the first doc that you could - as the last step of your test - verify that the correct document is open. It's need in some cases and would have to be re-implemented somehow. But that's limited effort
Jul 16 09:36:16 skotti	sunrayjogi: Agreed
Jul 16 09:36:51 sunrayjogi	All: Okay, want to finish this agenda item.....
Jul 16 09:37:45 oliverc	sunrayjogi, The global routine hclosedocument should be enhanced with DocumentCount() to avoid closing OOo
Jul 16 09:38:22 sunrayjogi	oliverc:  Agree...
Jul 16 09:38:57 skotti	oliverc: Does anyone know whether getDocumentCount really returns 0 on Backing-Window?
Jul 16 09:39:19 maho	So - I'm not against it, if you get rid of The first doc! we won't close whole office to proceed the other tests?
Jul 16 09:39:23 sunrayjogi	skotti:  It works fine. I have used it in my level-1 tests now.
Jul 16 09:39:38 skotti	sunrayjogi: Then i guess we're good to go-
Jul 16 09:39:46 sunrayjogi	maho:  you always should close the office if you start a new test (.bas)
Jul 16 09:39:52 sunrayjogi	skotti:  ok
Jul 16 09:39:58 maho	sunrayjogi : I understand :) thanks
Jul 16 09:40:26 ExLachs_home	maho, but between several testcases there would be the backingwindow
Jul 16 09:40:47 sunrayjogi	All: +1 or -1 for getting rif-off 'FirstDoc()' routine to OOo 2.4 which needs in some cases modifications to tests.
Jul 16 09:41:00 skotti	+1
Jul 16 09:41:03 sunrayjogi	+1
Jul 16 09:41:04 maho	+1 for getting rid-off
Jul 16 09:41:10 oliverc	+1
Jul 16 09:41:11 b_osi	+1
Jul 16 09:41:26 *	sunrayjogi 3.....
Jul 16 09:41:34 *	sunrayjogi 2.....
Jul 16 09:41:41 *	sunrayjogi 1.....
Jul 16 09:41:44 sunrayjogi	DONE

Can the waits be tuned in / shortened? (smo)

shifted to next meeting (smo not available)

Adding new scripts to first and topten (sanity check) (smo)

shifted to next meeting (smo not available)

Jul 16 09:45:10 *	sunrayjogi ....MEETING CLOSED....

Additional comments

Jul 16 09:46:06 ExLachs_home	maybe in the wiki the next meeting should be at the TOP not at the bottom
Personal tools