been using both for many years now, have never ran into issues or even been able to tell any difference in speed, let alone 10x
what I notice unfortunatly, is that I often miss search results with rg becuase I forget I need to pass the additional -i flag. this has shaped my perception of rg - extra focus on performance, sub-optimal ux
Whether smart case is enabled by default (as ag does) could easily go either way. Notably,.I think having it disabled by default is a better UX. But ripgrep does have a --smart-case flag, which you can add to an alias or a ripgrep config file. It also works more consistently than ag's smart case feature, which has bugs.
See my other comments about perf difference. And ag has several very critical bugs. And it's unmaintained.
> or even been able to tell any difference in speed
If you only search small amounts of data, then even a naive and very slow grep is likely just fine from a perf perspective.
I know this isn't the perfect place for a feature request, but I've recently found it quite annoying that ripgrep ignores .github directories because they're hidden. Maybe there are similar directories that people prefer included by default.
Then again, if these results suddenly pop up and people upgrade to a major version and expect backwards compatibility, the tradeoff might not be worth it.
> I know this isn't the perfect place for a feature request, but I've recently found it quite annoying that ripgrep ignores .github directories because they're hidden.
There have been some requests to whitelist certain directories by default. And it will absolutely never happen. It makes the semantics too complicated and even more surprising. Today, I can just say, "ripgrep respects gitignore, ignores hidden files and binary files." But if I started whitelisting specific things, that description would no longer be correct. I'd have to add them as exceptions when explaining ripgrep's default heuristics.
Plus, whitelisting a directory is extremely easy and simple. People seem to think `.ignore` files are only for ignoring. But they're not. They also let you un-ignore things without having to touch the CLI command.
> Then again, if these results suddenly pop up and people upgrade to a major version and expect backwards compatibility, the tradeoff might not be worth it.
I do use semver in ripgrep. That is, if I make a breaking change, then I bump the major version. Regardless, big breaking changes that aren't exceptionally well motivated are basically off the table. semver allows breaking changes to be made, but it works best in the context of libraries. For ripgrep, that has direct end users, semver doesn't really help much. Users will just experience their scripts breaking. Or commands they used to use, now don't work any more or do something different.
what I notice unfortunatly, is that I often miss search results with rg becuase I forget I need to pass the additional -i flag. this has shaped my perception of rg - extra focus on performance, sub-optimal ux