HN Explorer

14 comments

[ 0.23 ms ] story [ 9.2 ms ] thread
wyre
Thanks for developing scc!

Regarding LOCOMO I have a fork that calculates an estimate based on # of commits * price per task according to Artificial Analysis benchmarks for a given model, and includes a simple overhead multiplier to pad out the estimate.

boyter
You are welcome.

Id like to see that fork. Perhaps I can fold some of it in.

cnnrrss
Congratulations on the release boyter! scc has been a really useful tool for probing projects. Looking forward to the new features.
vivzkestrel
- saw this other post on your blog

- https://boyter.org/2010/08/mysql-backups-done-easily/

- is an sqldump actually a backup?

boyter
Its what I used to migrate to the 6 TB sqlite database... so yes? Depends on your definition of backup. In my case I wanted a way to restore in the case of data loss, so yes. Im sure some pedantic sql admin will disagree with me here.
ghoul2
if sqldump isn't a backup, what is?
vivzkestrel
they say pg_dump isnt a backup hence my question
aslakhellesoy
Very interesting. This reminds me a lot of the CRAP metric - a combination of cyclomatic complexity and code coverage.

It would be intersting to combine your «attention» metric with CRAP.

boyter
Not one I have come across. Now in my list of reading.
jmfn
Neat tool. Wondering if this has a place helping to gate PR’a that should require human review.
boyter
Probably? You could look at hotspots using it and determine a human in the loop must happen.

If you do hook this up please let me know, id love to write about it.

othmanosx
Interesting, I built https://pyor.review to fix this exact issue, though it uses AI to categorize the files and groups them by complexity to speed up the review process because most AI generated code is noise, especially migration PRs. Curious to know if this could be done in a deterministic way without the help of AI
boyter
Thats what this should be able to do for you. Feel free to hook it in and see if it works. If so please let me know!