Clojurians Log v2

I think it might be better to remove the follow-links behavior in delete-tree as this seems uncommon. Not even rm -rf supports it.

https://stackoverflow.com/questions/2977536/deleting-a-folder-that-contains-symlinks#:~:text=No.,it%20will%20simply%20remove%20them.

Or I don't know. I'll ask a second opinion on this. There might be use cases, but it seems dangerous

https://clojurians.slack.com/archives/C03RZGPG3/p1612689372122100

fwiw, I have a bb script that does recursive delete, and I have it check for symlinks first and fail if it finds any. This might have been before I realized a delete only deletes the link... just the same I was obviously feeling very cautious about the danger accidentally over-deleting something.

right. so same as rm -rf

not including the option anymore

cool, good thinking

^ fun lein imitation script based on deps.edn

klein 😂

I am writing a script that is intending to process multiple project’s deps.edn, install the deps, and generate the classpath. Ideally I would to use babashka.deps or babashka.clojure but how would I isolate them? How would I run it in such a way where .cpcache and such are generated according to the project’s respective directory?

@jayzawrotny You mean something like babashka.deps/add-deps but call it multiple times and get only the resulting classpath for each file?

Correct

Now that I think about it I don’t think babashka.deps or babashka.clojure is the best tool for the job on this one. Perhaps I’m better off shelling out to bb --clojure to process those deps as a separate process in their respective project directories?

Calling add-deps does what it says: it only adds deps to the classpath, it doesn't replace the previous classpath. However, you can hack around this by storing the previous classpath and use str/replace to remove it from the next one. This is relying on the impl detail that the classpath is only concatenated currently and there is no de-duplication whatsoever.

yeah, you can shell out using babashka.process or clojure.java.shell and set the :dir argument to start the processes from a different dir

I think that would be the most reliable

And if I combine that with the --clojure flag it should work on windows?

yes, it should

this is one of the reasons I included this, since clojure on Windows is a bit messy with powershell and shelling out and such

Good idea!

The fs package came at a perfect time too, I didn’t even think about how platform specific my path separators were 😅

:)