| *** haleyb|out is now known as haleyb | 13:00 | |
| sean-k-mooney | o/ am not sure if anyone else has had isseu with past.opendev.org but when i try to past baseiclly anything i get 502 proxy erros back form it | 16:19 |
|---|---|---|
| sean-k-mooney | that not just a me thing right? | 16:19 |
| clarkb | https://paste.opendev.org/show/bvBWt51BbyRxrzkA0yFW/ just worked for me. I suspect that the content of your paste is at least part of the problem | 16:19 |
| clarkb | like maybe it is trying to use a captcha and failing? I can look to see if logs have any info shortly | 16:20 |
| sean-k-mooney | hum its just my shell output | 16:20 |
| sean-k-mooney | let me put it somewhere else for you do see | 16:20 |
| sean-k-mooney | https://termbin.com/afim | 16:21 |
| sean-k-mooney | ok ya... | 16:22 |
| sean-k-mooney | https://paste.opendev.org/show/b4aSijWMYkvrRjnESSoW/ worked | 16:22 |
| clarkb | it doesn't like your ➜ | 16:22 |
| sean-k-mooney | oh ok | 16:22 |
| clarkb | that appears to be a 4 byte utf8 char. I'm guessing the db is 3 byte only | 16:22 |
| sean-k-mooney | i dont think that is new but i have been using kde for the last 3 weeks with a diffent terminal | 16:23 |
| sean-k-mooney | so maybe that has changed | 16:23 |
| clarkb | we haven't touched the db for that in years which also likely explains why it is 3 byte utf8 | 16:23 |
| fungi | we ought to be able to switch it, but i thought we had changed all our systems to 4-byte encoding default for mariadb. maybe we missed paste | 16:23 |
| clarkb | fungi: ya modern mariadb does utf4 by default I think. But the old dbs when exported and imported keep the old utf8mb3 or whatever it is | 16:24 |
| clarkb | we probably need to update lodgeit and then do a db migration | 16:24 |
| clarkb | hrm actually may be a different char. I'm grabbing the hex code from the log and will see what python translates that to | 16:25 |
| sean-k-mooney | i can just update my starship config to not put that in ps1 | 16:25 |
| sean-k-mooney | or whaterver | 16:25 |
| sean-k-mooney | i think i updated the version i have installed recently | 16:25 |
| clarkb | 🐍 | 16:26 |
| sean-k-mooney | feel free to fix it on the db side to but since i know the causei can work around it. i was wondering if one of the emogis was the issue | 16:26 |
| clarkb | I think it was ^ not the arrow | 16:26 |
| fungi | my terminal doesn't have that glyph | 16:26 |
| fungi | what does unicodedata.name() say it is? | 16:26 |
| clarkb | fungi: b'\xF0\x9F\x90\x8D' | 16:26 |
| clarkb | 'SNAKE' | 16:27 |
| fungi | >>> unicodedata.name(b'\xF0\x9F\x90\x8D'.decode('utf-8')) | 16:27 |
| fungi | 'SNAKE' | 16:27 |
| clarkb | "Incorrect string value: '\\xF0\\x9F\\x90\\x8D v...' for column `lodgeit`.`pastes`.`code` at row 1" | 16:27 |
| fungi | indeed | 16:27 |
| sean-k-mooney | oh | 16:27 |
| sean-k-mooney | its not happy with the python sybol | 16:27 |
| clarkb | that error comes from pymysql | 16:27 |
| clarkb | and is mariadb error 1266 | 16:28 |
| clarkb | *1366 | 16:28 |
| fungi | right, so the db connection in lodgeit is probably still using utf8mb3 | 16:28 |
| sean-k-mooney | so that has been in my modeline for a number of years but they may have changed the encodeing or unicode charter at some point | 16:28 |
| clarkb | it is possible older versions simply truncated the string. I seem to recall etherpad did something similar | 16:31 |
| clarkb | but as we've updated things to keep up with python releases the error is less forgiving | 16:31 |
| sean-k-mooney | ya thats posisbel i can just avoid copying the prompt line for now | 16:32 |
| clarkb | its probably the sysadmin in me but I avoid all non ascii characters as much as possible for things I rely on | 16:32 |
| sean-k-mooney | its nice locally to have the extra info but if im sharing command output its not relly needed | 16:32 |
| clarkb | blowing up on utf8 is a common issue | 16:32 |
| fungi | there are a bunch of snake codepoints, so might have been using a different one previously? | 16:33 |
| clarkb | the matrix gerritbot container broke because haskell logging decided to do utf8 checkmarks by default then the container image they buitl it in didn't have a utf8 locale present... as a non mysql example | 16:33 |
| sean-k-mooney | ya given how lon utf8mb3 has been deprecrated i expect most install to be usign 4 byte now but there are a number of possibliteis | 16:33 |
| fungi | there are 5 different egyptian snake heiroglyphs alone, all of which are 3-byte codepoints | 16:34 |
| sean-k-mooney | i dont think this is imporant enough to consume your time | 16:34 |
| clarkb | I think rhel 7 also doesn't do C.utf8 | 16:34 |
| fungi | still? | 16:34 |
| fungi | oh, right, *7* | 16:34 |
| fungi | i keep LC_CTYPE, LC_COLLATE and LC_MEASUREMENT all set to C.UTF-8 on my personal systems | 16:36 |
| clarkb | sean-k-mooney: fwiw on services like gerrit and etherpad we do test 4 byte ut8 works | 16:36 |
| clarkb | though the gerrit test may have been a one off | 16:36 |
| clarkb | I pushed a change with a ton of different characters to ensure they rendered (and they mostly do) | 16:36 |
| sean-k-mooney | clarkb: before i fix this up locally to use somethin else | 16:37 |
| sean-k-mooney | me posting variation of my modelien until it work wont break anything right | 16:37 |
| fungi | we first experienced the issue during a design summit when someone crashed the therpad server with a snowman | 16:37 |
| fungi | er, the etherpad server | 16:37 |
| clarkb | sean-k-mooney: ya I think that is fine. Mariadb is just rejecting the data in the first place and lodgeit bubbles that up as a 502 apparently | 16:38 |
| sean-k-mooney | taht impressive | 16:38 |
| clarkb | I don't think it is impacting anything in the db proper | 16:38 |
| sean-k-mooney | i dont tend to deep dive into alot of customisation but i wonder if starship has an acsii mode | 16:39 |
| clarkb | https://review.opendev.org/c/opendev/system-config/+/900379 the gerrit check | 16:39 |
| clarkb | and ya its not automated but instead just somethign we can check periodically | 16:39 |
| sean-k-mooney | that would be nice but if not ill jsut tweak the common thing i use it for | 16:39 |
| sean-k-mooney | liek git branch and virtual env to not have unicode | 16:39 |
| clarkb | the little tab indicators overlap with the chars in some cases which sin't great but it doesn't explode in gerrit either | 16:40 |
| clarkb | wget uses ""s that aren't ascii either | 16:40 |
| fungi | yeah, mixing "east asian width" characters into terminal content often leads to alignment issues for me | 16:41 |
| clarkb | https://opendev.org/opendev/system-config/commit/73d2d784e42fa2e04bd513857ca5f2ed62c4cc1e and here is gitea rendering that same file | 16:42 |
| fungi | not to mention line-wrapping routines have to take them into account as consuming additional columns | 16:42 |
| fungi | funny that it considers line 15 to contain ambiguous characters, but not line 5 | 16:43 |
| clarkb | if you hover the red outlined glyph it tells you what it is complaining about which is nice | 16:44 |
| fungi | it doesn't render the red outline for me, just the warning triangle | 16:45 |
| fungi | maybe that's a difference with its "dark mode" | 16:45 |
| clarkb | ya I'm in light mode | 16:45 |
| fungi | or maybe it's just a firefox issue, who knows | 16:45 |
| clarkb | is the second glyph on the line the U | 16:45 |
| fungi | yah, i still get the pop-up, but i had to hover over each character in turn to find the right one | 16:46 |
| clarkb | U+222A can be confused with U+0055 | 16:46 |
| fungi | 0x55 is ascii "U" | 16:46 |
| clarkb | maybe it only considers things that can be confused with ascii because of problems with say urls? | 16:47 |
| clarkb | ya | 16:47 |
| clarkb | the X variants are maybe ok because they don't look like an ascii X? | 16:47 |
| fungi | i guess it depends on the font, ❌ and X look pretty similar in my terminal, as do ✗ and x | 16:47 |
| clarkb | ∪ and U render sufficiently different in my terminal font (source code pro) but look similar in the web font whatever is being used by gitea | 16:47 |
| fungi | yeah, maybe that's it. i can easily distinguish them all side-by-side in my terminal fwiw, but similarities in the gitea web interface are what matters for the warning, sure | 16:48 |
| fungi | gerrit has a similar warning, i seem to recall, but that change isn't tripping it | 16:49 |
| sean-k-mooney | ok well https://paste.opendev.org/show/bh6euk9Dxw21pbJEigb6/ does not break it so i guess im good | 16:50 |
| sean-k-mooney | fungi: i remember the first time stephen finucan added an emogi to a commit message and gerrit was not a fan | 16:51 |
| sean-k-mooney | that hasnt been a probelm in a very long time | 16:51 |
| clarkb | ya I think the specific case I was trying to test was for the code editor. The RO rednering has been good for a long time | 16:52 |
| sean-k-mooney | i dont tend to use emigi personly. i use emoticons :) but that about it | 16:53 |
| fungi | same here! | 16:53 |
| sean-k-mooney | so i rearly hit issue liek this | 16:53 |
| sean-k-mooney | clarkb: fungi i know we do a lot in the public but for future refeicne i assuem the db log is not public right. i.e. i coudl not have debugged this myself if i knew where to look | 16:58 |
| frickler | sean-k-mooney: yes that's only local on the server in this case | 16:59 |
| sean-k-mooney | ack cool | 16:59 |
| sean-k-mooney | im not asking to change that just wnat to avoid wasting other time if i can fix it myself | 16:59 |
| clarkb | yup we're generally careful about those logs because software is often bad about hiding secret info like passwords | 16:59 |
| sean-k-mooney | yep understandable | 17:00 |
| clarkb | but we're happy to dig into them and share when appropriate. I shared what i believed to be the relevant portion above. Happy to include more if people want it | 17:00 |
| fungi | even openstack itself is not great about hiding sensitive information when logging | 17:00 |
| sean-k-mooney | epically if you dont use oslo.config | 17:01 |
| fungi | or turn on debug | 17:01 |
| sean-k-mooney | but if you ever see somethign that shoudl not be there for nova let me knwo | 17:01 |
| sean-k-mooney | well no we try to make debug valid to have in proudction | 17:01 |
| fungi | yeah, we consider those security hardening opportunity bugs | 17:01 |
| sean-k-mooney | without passwards | 17:01 |
| sean-k-mooney | yep | 17:02 |
| fungi | it's gotten a lot better in recent years | 17:02 |
| *** benj_2 is now known as benj_ | 17:39 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!