[13:21:11] Hey all I’m going to cancel the retro today. With Aaron leaving and Andy out there is less value. KevinBazira halfak haksoat [13:28:14] OK sounds reasonable. [13:29:05] chrisalbon, what do you think about moving BS time up into that slot since we're going to need to do tuning session stuff during the usual time? [13:29:34] That works [13:31:11] Moved! [13:31:19] haksoat, kevinbazira ^ [15:02:51] Thanks halfak. [15:05:04] :) [15:26:24] chtnnh, run "python setup.py install" in the base directory. [15:26:42] sure halfak just a minute [15:28:49] running the make again [15:34:54] built successfully! halfak [15:36:01] Awesome! Get it in the PR. Let's look at those stats :) [15:40:09] on it [15:44:03] okay there seems to be an issue with the 2fa rn halfak it returns a 403 [15:46:51] Dang. Where are you getting the 403? When trying to push? [15:47:01] Are you using your dev key? [15:47:18] 403 is a permission issue that you might get for typing your password rather than the dev key :P [15:49:54] yeah im using my personal access token [15:50:20] its doing the weird thing like last time remember? access to chtnnh/articlequality denied to chtnnh [15:50:36] so its recognizing me for chtnnh but its not letting me commit for some weird reason [16:52:59] Hello halfak [16:53:16] The build for the fresh 10k vectors finished [16:53:18] https://github.com/wikimedia/drafttopic/pull/53/commits/51cbfc612a3ccab937c82ba4917fc879f5a591c8 [16:53:22] chtnnh, weird! That must be github. [16:53:31] Is it still messing up? [16:53:37] yes no progress [16:53:38] pr_auc dropped though, but a minimal drop this time [16:53:38] haksoat, checking [16:53:40] tried stack overflow [16:53:54] chtnnh, can you make a paste of the error? [16:54:08] just a second [16:54:23] haksoat, am I reading this right that the fresh 10k worked better? [16:54:44] But still not quote as good as the old 100k? [16:55:19] Yes [16:55:22] That [16:55:23] https://gist.github.com/chtnnh/07f041e7b88c370b22795810ee55e8ad halfak [16:55:25] right [16:55:43] haksoat, let's stick with the 10k. I think these results are good enough. [16:56:57] So now for merging, we should merge https://github.com/wikimedia/drafttopic/pull/52 and then do a rebase on master in your branch and then merge it. [16:57:01] How does that sound? [16:58:43] Sounds good [16:59:28] OK merged it. Pull changes to master and rebase then force push to your PR. Let me know and I'll do the final review :) [17:06:13] Cool [17:44:30] halfak: I just pushed. Tests still running though [20:04:42] * halfak goes to see what is taking travis so long. [20:05:28] * halfak restarts the build [20:21:02] wikimedia/editquality#755 (enwiki_reverted_test - f28ed31 : Aaron Halfaker): The build failed. https://travis-ci.org/wikimedia/editquality/builds/708114041 [21:10:28] halfak: o/ I had a small question - can python-mwxml directly work with gzipped files? or do i need to extract them? [21:10:44] It works directly with gz, bz2, and 7z. [21:10:47] :D [21:11:09] thanks! :S [21:11:12] * :D [21:11:38] \o/ [21:12:34] I just merged haksoat's work. [21:12:55] Looks like we might be able to do a real ORES deployment if you're up for it, chrisalbon. [21:12:58] :D [21:13:15] Heck yeah [21:13:31] I'm really excited to see how much of a bit we took out of memory usage. [21:13:47] We might be able to increase the number of topic models we have deployed substantially! [21:26:26] Heading out for the day. Have a good evening, folks. [22:21:10] I'm trying to read the .gz dump using mwxl, but i'm getting a decode error - "UnicodeDecodeError: 'utf-8' codec can't decode byte 0x8b in position 1: invalid start byte" [22:21:24] using it as described here - https://github.com/mediawiki-utilities/python-mwxml [22:21:46] dump file i'm using is - enwiki-20200701-stub-meta-history.xml.gz [23:16:36] silly question, but did you try with the decompressed file? [23:16:55] ah, halfak said it should work fine