I'm very glad to see work being done to remove some of our redundant layout test results: http://trac.webkit.org/changeset/66124 I'm curious what the status of that work is. Do we expect all duplicates are gone now? or just chromium's? -eric
I removed all of chromium's duplicate results, which was a bulk of the duplicates. I think there were a total of about 670 duplicate files and I deleted about 500 or so. The remaining duplicates are mostly cases where the same result file is in mac and mac-leopard. I can delete those on Monday. There aren't as many duplicates as I thought. 670 is a lot, but not compared to the 90,000+ files in LayoutTests. On Sat, Aug 28, 2010 at 10:47 AM, Eric Seidel <eric@webkit.org> wrote:
I'm very glad to see work being done to remove some of our redundant layout test results: http://trac.webkit.org/changeset/66124
I'm curious what the status of that work is. Do we expect all duplicates are gone now? or just chromium's?
-eric
On Sat, Aug 28, 2010 at 10:47 AM, Eric Seidel <eric@webkit.org> wrote:
I'm very glad to see work being done to remove some of our redundant layout test results: http://trac.webkit.org/changeset/66124
I'm curious what the status of that work is. Do we expect all duplicates are gone now? or just chromium's?
The script as written was Chromium-agnostic (it uses the layout test fallback lists from the normal WebKit scripts). I think Tony just wasn't completely confident it was going to do the right thing so he was deleting only a subdirectory at a time. (Last week he found and fixed a bug in it: when you find duplicate results for platform A that falls back on platform C, you need to consider intermediate platform B's results.) If you have a WebKit build and some time to spare, you can try deleting the redundant tests for your platform that it suggests deleting and seeing if your platform still passes the tests to verify before checking in the deletes.
participants (3)
-
Eric Seidel
-
Evan Martin
-
Tony Chang