| 9 Jul 2025 |
 nh2 | jonhermansen: yes, copying it to your own location also ensures that given it works for hexa , we can exclude that some of us got different bytes delivered by that URL | 01:36:42 |
 jonhermansen | This URL seems to work fine, but it has different contents https://github.com/SeleniumHQ/selenium/releases/download/selenium-3.141.0/selenium-server-standalone-3.141.0.jar | 01:37:36 |
| jonhermansen | ~ curl -s https://selenium-release.storage.googleapis.com/3.141/selenium-server-standalone-3.141.59.jar | sha256sum
acf71b77d1b66b55db6fb0bed6d8bae2bbd481311bcbedfeff472c0d15e8f3cb -
| 01:38:31 |
| jonhermansen | ~ curl -s https://selenium-release.storage.googleapis.com/3.141/selenium-server-standalone-3.141.59.jar | file -
/dev/stdin: Zip archive data, at least v2.0 to extract, compression method=deflate
~ curl -s https://selenium-release.storage.googleapis.com/3.141/selenium-server-standalone-3.141.59.jar | sha256sum
acf71b77d1b66b55db6fb0bed6d8bae2bbd481311bcbedfeff472c0d15e8f3cb -
| 01:39:11 |
 dramforever | huh, i think i found it: the failing url returns an http header content-encoding: None | 01:40:32 |
| dramforever | that's where None has been coming from | 01:40:42 |
| jonhermansen | LOL | 01:40:49 |
| jonhermansen | can you see verbose curl info when running Nix? | 01:41:10 |
| dramforever | doesn't seem like it | 01:41:54 |
| jonhermansen | but someone else ran the command on another system and got a different result? is the result also dependend on the sqlite DB state? | 01:42:08 |
| jonhermansen | * but someone else ran the command on another system and got a different result? is the result also dependent on the sqlite DB state? | 01:42:17 |
| nh2 | looks like that message should be improved to make that clear | 01:42:26 |
| dramforever | unlikely, given that this thing fails at the http download stage | 01:42:29 |
| jonhermansen | I am a complete Nix noob so I'm learning as I go, if I ask a dumb question, sorry | 01:42:31 |
| dramforever | please ask more dumb questions | 01:43:03 |
| jonhermansen | Ok, duh, the .jar file is coming from Google in this instance | 01:43:07 |
| dramforever | so the http server is sending a bad header | 01:43:30 |
| dramforever | and we don't handle that particularly gracefully | 01:43:37 |
| dramforever | it looks like in TransferItem::writeCallback we call makeDecompressionSink | 01:44:14 |
| dramforever | passing the encoding straight from the http header to libarchive | 01:44:32 |
| dramforever | and libarchive goes "i literally have no idea what None is" | 01:44:41 |
| jonhermansen | How big is the buffer that stores the encoding string? | 01:45:09 |
| dramforever | normally it should only give what we send as Accept-Encoding | 01:45:21 |
| jonhermansen | This is libcurl right, I guess nix includes in | 01:45:22 |
| nh2 | Repro: curl --silent --head 'https://selenium-release.storage.googleapis.com/3.141/selenium-server-standalone-3.141.59.jar' | grep -i content-encoding
I get
x-goog-stored-content-encoding: None
content-encoding: None
hexa do you get that also, given it worked for you?
| 01:45:27 |
| jonhermansen | * This is libcurl right, I guess nix includes it | 01:45:31 |
| jonhermansen | yup | 01:45:50 |
 hexa | yes | 01:45:51 |
| dramforever | i'm not sure why but i think we handle decompression ourselves | 01:45:51 |
| jonhermansen | x-goog-stored-content-encoding: None | 01:45:53 |
