Closed ppigazzini closed 4 years ago
These new builds have been working fine for ten days on multi variant fishtest and on DEV (with my workers).
cutechess-cli v1.2.0 has updated the output:
Started game 19 of 20 (base vs test)
Finished game 19 (base vs test): 0-1 {Black wins by adjudication}
Score of base vs test: 4 - 10 - 5 [0.342] 19
... base playing White: 3 - 4 - 3 [0.450] 10
... base playing Black: 1 - 6 - 2 [0.222] 9
... White vs Black: 9 - 5 - 5 [0.605] 19
Elo difference: -113.6 +/- 149.6, LOS: 5.4 %, DrawRatio: 26.3 %
Started game 20 of 20 (test vs base)
Finished game 20 (test vs base): 1/2-1/2 {Draw by adjudication}
Score of base vs test: 4 - 10 - 6 [0.350] 20
... base playing White: 3 - 4 - 3 [0.450] 10
... base playing Black: 1 - 6 - 3 [0.250] 10
... White vs Black: 9 - 5 - 6 [0.600] 20
Elo difference: -107.5 +/- 139.5, LOS: 5.4 %, DrawRatio: 30.0 %
Finished match
previous cutechess-cli:
Started game 19 of 20 (base vs test)
Finished game 19 (base vs test): 0-1 {Black wins by adjudication}
Score of base vs test: 7 - 7 - 5 [0.500] 19
Elo difference: 0.0 +/- 141.2, LOS: 50.0 %, DrawRatio: 26.3 %
Started game 20 of 20 (test vs base)
Finished game 20 (test vs base): 1-0 {White wins by adjudication}
Score of base vs test: 7 - 8 - 5 [0.475] 20
Elo difference: -17.4 +/- 138.8, LOS: 39.8 %, DrawRatio: 25.0 %
Finished match
do we need to make changes to the worker to deal with this changed output ?
IMO no, better to ask to @vdbergh too.
I'll have a look tonight.
@vondele your PR https://github.com/cutechess/cutechess/pull/551 was merged, cutechess-cli version is raised, so IIRC we should be able to fix https://github.com/glinscott/fishtest/issues/478
yes, I agree we should be able to deal with https://github.com/glinscott/fishtest/issues/478 now. I know what needs to be done to the worker code, in principle, but this also requires passing some additional info from the server.. I'll see if I can figure this out.
I wait til @vdbergh confirms we can use this without changes, before I merge this PR.
This is a low priority PR, useful only to work on glinscott/fishtest#478 Anyway my workers on DEV are already using the new cutechess-cli with:
git clone --single-branch --branch master https://github.com/glinscott/fishtest.git
cd ${test_folder}/fishtest
git config user.email "you@example.com"
git config user.name "your_name"
# add here the upstream branches to be tested
git remote add new-upstream https://github.com/ppigazzini/fishtest
git pull --no-edit new-upstream TEST_ppigazzini_books
@ppigazzini see https://github.com/cutechess/cutechess/issues/616 I think standard chess is not affected, but I'm not 100% sure. (I've asked in the issue).
@vondele subscribed, thank you.
Cutechess binaries built from v1.2.0 cutechess/cutechess@3d1ac5b7 using these instructions.
All files tested clean with VirusTotal
cutechess-cli-win.zip
cutechess-cli-linux-64bit.zip
cutechess-cli-linux-32bit.zip