Hacker News new | ask | show | jobs
by Oddskar 1650 days ago
I still think the changes made to only index repos that had activity during the last year [1] was the wrong call, and potentially makes code search dangerous for those that do not understand this.

At least for me a common use case for org-wide code searches is to answer questions like "is the library x still used somewhere?"

To not have this include old potentially low-traffic repos is a very bad thing. I don't understand why they would do this for enterprise customers. Like we would pay extra to not have it be this way.

[1] https://github.blog/changelog/2020-12-17-changes-to-code-sea...

3 comments

Unfortunately I lost trust in GitHub search a while ago. Can't find files, can't find an exact string reference. This can be dangerous in certain situations like you said. Hopefully I can start to gain trust in the new version when released but I don't really see myself using it again.
Same here. GitHub search sucks for authoritative, governance use cases. I got burned a few times and now just pull everything down myself and search it.

I think there’s code search companies but they are too expensive for me and I suppose some people really value it more highly.

Comically, I need the old Google search appliance and just treat it as a web source for lots of my questions (who is using log4j, etc).

> I need the old Google search appliance

Or Google Code Search! Which turned out to be Russ Cox's platform for developing RE2.

Google Code Search was simply amazing. Killing Google Code Search really underlined how Google had given up on it's pro-social agenda. I mean, here was a best-in-class and as yet unmatched service, neatly tailored to both the needs of developers and to Google's expertise and operations, and they killed it. Why? Because it wasn't going to make any money? Was there ever an expectation that it would?

The death of Google Books, when they removed countless scanned items, most of which were clearly out of copyright, was also painful. But I could understand how dealing with the barrage of copyright lawsuits was simply a bridge too far for Google. But Code Search? Why, oh, why!? :(

> just pull everything down myself and search it.

`grep -r` or `git-all grep` or something better?

I'm not who you replied to but I use ag and it's pretty fast (faster than grep I think)

https://github.com/ggreer/the_silver_searcher

Yes this seems like a bad move to me too. Plenty of repos still useful may not change in a year. But still be relevant.
It won't search forked repositories either, which is a pain when trying to find something - you're better off pulling and grepping.

(If you have forked a long-dead project and are working on it Github support can "decouple" it from the original and then you can search it.)