486062 - Clean out dead things below layout/html/tests/

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Phil Ringnalda (:philor)]

Attachment: Remove 'em

Most of this is removal based on my understanding of the prior generations of semi-automated tests: before the current generation was one that was run by building the layout-debug extension, then telling it to run a baseline on the things in layout/html/tests/block/rtest.lst, make a change, tell it to run verify on the same list, which still works so I'm leaving it; before that was a half-generation of rtest.sh/runtests.sh which ported the first generation to Linux, and thus required the now-long-gone viewer, and the first generation was run from rtest.bat, on Windows, with a srcdir build, when viewer existed, and with the source tree available at s:\mozilla\. So unless I missed some, I removed all file_list.txt/rtest.sh/runtests.sh/rtest.bat without mentioning them.

From the top (in mxr-sorting):

* CellMapTest/CellMapTest.cpp has never been built by default; because it expects to include "..\..\src\nsCellMap.h" I expect it has never been built at all.

* block/
  * dom/ isn't included in rtest.lst
  * interactive/ is interactive, but saying "change the window size" without saying what's expected makes it awfully unsatisfactorily interactive
  * dummy.html is only included for unexplained reasons in bug/file_list_slow.txt

* formctrls/interactive/ - yeah, manual tests which require you to re-read and re-understand the bug every time you run them, to know what success looks like, probably not such a great idea

* frameset/ - apparently when you were doing rtest.lst, this didn't excite you, so it's been dead for five years or so

* printer/ - despite this being dead without viewer, it would be a treasure trove for a hypothetical printing QA person, so I couldn't bring myself to kill it

* style/base/ - you must have felt that other things did a good enough job with base style, since this isn't in rtest.lst

* table/ - stepping lightly here, since Bernd retained far more interest in these tests than anyone else for far longer, so I just removed file_list.txt, rtests.bat, and...

* table/TableContentTest/TableContentTest.cpp - neither built, nor likely to be buildable

* tags/ - it's cute as can be that we had tests for only six HTML tags, but included within those six both plaintext and xmp.

Comment 1

[Bernd]

I have no problem with removing them as we do have the files on previous branches. Although I would have prefered a conversion of them before removal like I did with table/dom but thats to much work and requires fruitless digging into old bugs.

Comment 2

[Phil Ringnalda (:philor)]

Well, sure, if it's what the two of you want I can just hg rm layout/html/* instead.

That does have at least one advantage, in that people won't think we have test coverage of something when we only have vague manual coverage that's maybe never actually run at all (if I was in a hurry, I'd probably say from http://mxr.mozilla.org/mozilla-central/search?string=list-style-type%3A+lower-alpha&find=test that we have test coverage of list-style-type: lower-alpha), but also the disadvantage that it's vastly less likely that someone will go back and port tests from 1.9 than that they'll port tests which are hanging around in their 2.5 tree.

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Adding bz and roc to make sure they're ok with removing these as well.

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Yes, lets!

Comment 5

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 370139 [details] [diff] [review]
Remove 'em

ok, r+sr=dbaron; sorry for the really long delay here.

Comment 6

[Phil Ringnalda (:philor)]

No problem with delay: it's not like I'm going to get bitrotted ;)

http://hg.mozilla.org/mozilla-central/rev/f662f710fead

Comment 7

[Boris Zbarsky [:bzbarsky]]

This is fine with me too, obviously.

I wonder whether it'd be possible to set in-testsuite? on the bug#s mentioned in those tests somehow, though...

Comment 8

[Phil Ringnalda (:philor)]

Only manually - once I noticed that they looked to be from a "kw:testcase or sw:testcase" query, I headed off to Change Several Bugs At Once, only to find yet again that you can't set flags that way.