Marcel Bargull

> Wow, talk about a deep rabbit hole 🐰 Good thing is, that I was able to pin point this very quickly. I didn't even do any profiling, just a...

Addendum: > Interestingly, with the `classic` solver we seem to take slightly different code paths with much less impact. This is only happened because `--solver=classic` cheated by using `current_repodata.json`. Running...

> > ulimit -Sv 1000000 && > > Oooh, that's a nice tip! It's a convenient thing because one can just add it on top of everything. But it's also...

> There a few places where we are calling `isEnabledFor` that are worth revisiting: Probably; but that would be orthogonal to this issue (i.e., should go into a separate issue)....

Thanks for working on this @zklaus! > I have been investigating this a bit. To sum up my understanding so far: \[...\] Seems about right :thumbsup:. My understanding in other...

> > We could rethink logging as a whole > Hard no, at least based on the presented reasons above. Of course there are different libraries that are better suited,...

I've proposed gh-13628 as a fix for the excess resource usage due to this issue for now. We could/should include that PR in the `24.3.x` milestone. It is of course...

> So if you have `$PREFIX/lib/library.so` which symlinks to `$PREFIX/lib/nested/deeply/library.so` you will compute an RPATH entry of `$ORIGIN/../../`. When `$PREFIX/lib/library.so` is loaded it will now search for libraries outside the...

Thanks for the details. I have to admit I was a bit oblivious about the issues we can run into with this -- thanks for making me aware; much appreciated!...

Ah, I forgot to comment on this one, sorry. From what it looks like (hard to tell exactly because of fluctuating run times), it kind of cancels each other out....