Hitler reacts to Conda failure
6 views • 7/31/2025
Reaction to common conda install errors, familiar to many in the field.
00:00 - 00:03 | The conda install failed again, sir. |
00:04 - 00:05 | They created a fresh conda environment. |
00:05 - 00:07 | Ran `conda install -c bioconda samtools`. |
00:08 - 00:12 | The solver couldn't resolve the dependencies. |
00:12 - 00:15 | We thought switching to Mamba would fix it. |
00:17 - 00:19 | Have you also tried making a new env |
00:19 - 00:21 | or trying to install it with Mamba? |
00:24 - 00:26 | Mein Führer... |
00:27 - 00:28 | Mamba— |
00:31 - 00:33 | Mamba failed instantly, and making a new env gave the same repeated error |
00:34 - 00:36 | Same error, fewer seconds |
00:53 - 00:58 | Everyone who still codes in R, leave the room. Keitel, Jodl, Krebs, Burgdorf, stay here. |
01:13 - 01:15 | I followed every step! |
01:15 - 01:17 | I made a clean environment! |
01:18 - 01:23 | I used conda-forge and bioconda! |
01:25 - 01:28 | Updated Conda! Cleaned the cache! |
01:29 - 01:31 | Mamba was supposed to solve everything in seconds! |
01:31 - 01:34 | Instead it failed instantly—just like Conda! |
01:34 - 01:37 | I even tried pip from the GitHub repo! |
01:37 - 01:40 | That failed too! |
01:40 - 01:42 | Mein Führer, maybe we can try installing it inside Docker... |
01:42 - 01:46 | They said Conda makes this easy! |
01:46 - 01:48 | Mein Führer, many labs use containers now to manage issues like this |
01:48 - 01:52 | What’s the point of going through all this trouble? We may as well just install every dependency... |
01:53 - 01:54 | One by one! |
01:56 - 01:57 | This isn’t reproducible science- |
01:57 - 02:00 | This is psychological warfare! |
02:00 - 02:03 | I just wanted to preprocess a BAM file. |
02:04 - 02:08 | Now the Jupyter kernel won’t even start. |
02:08 - 02:13 | I launched it once and now it's stuck in “Dead” forever. |
02:14 - 02:16 | Every fix I try just makes it worse. |
02:17 - 02:21 | I can't even remember which environment I'm in anymore! |
02:27 - 02:29 | I’ve made five environments named “test”! |
02:30 - 02:34 | They all broke in slightly different ways. |
02:34 - 02:36 | All I want is to process some FASTQ files! |
02:41 - 02:42 | From someone else's paper! |
02:43 - 02:47 | Because they were too lazy to upload a DESeq2 table. |
02:48 - 02:53 | No raw counts. No supplementary scripts. |
02:54 - 02:56 | Just a GEO accession and a broken link! |
02:56 - 02:59 | “Data available upon reasonable request” my ass. |
03:00 - 03:02 | I’m not asking for their raw alignments— |
03:04 - 03:07 | It's okay, we're experimentalists, we don't have to deal with this shit |
03:14 - 03:16 | Now I’ve spent six hours fixing conda just to process by STAR alignments |
03:19 - 03:23 | I haven’t even processed a single read yet. |
03:25 - 03:26 | And I think I forgot to trim the adapters. |
03:31 - 03:33 | Every time I fix one thing, three more break. |
03:40 - 03:46 | This was supposed to be a simple reanalysis. |
03:46 - 03:49 | Why did I ever agree to do this? |
03:53 - 03:56 | I should have went into pure CS. |
No comments yet.