I guess if you use a rigid fingering system. I never bothered to learn that back in elementary school, so my fingering is ad hoc based on what I'm typing. To type "rg" (using QWERTY), I'd just bring my middle finger up to R while using my index finger for G. This would probably be a little slower than "ag" because my hand is more likely to already be in place to type the latter without movement ("home row"), but not as slow as reusing a finger would be. It's not something I'd have to think about; this is already what I do whenever I have to type "rg" as part of a word.
I'm curious whether an ad-hoc approach is more or less efficient overall. Fingering customized per word clearly has the potential to optimize finger movement, but my error rate is relatively high - mostly timing-related - which might be exacerbated by an ad-hoc system because there are more (and more complex) unfamiliar transitions between words.
Anyway, I upvoted you for mentioning typing. It might seem trivial - well, if you use a lot of custom aliases, it is trivial - but if a command runs fast enough (and ag is already very fast on small source trees), the time spent typing its name can become a significant bottleneck. The author of Pijul, for instance, a version control system meant to compete with Git, seems not to recognize this... the command is 'pijul', which is essentially impossible to type on QWERTY without reusing a finger at least once.
I'm not super rigid with all keys but I do seem to have "R with index finger" ingrained. R with the middle finger does work nicely in this case if I think about it and force myself to do it.
I guess if you use a rigid fingering system. I never bothered to learn that back in elementary school, so my fingering is ad hoc based on what I'm typing. To type "rg" (using QWERTY), I'd just bring my middle finger up to R while using my index finger for G. This would probably be a little slower than "ag" because my hand is more likely to already be in place to type the latter without movement ("home row"), but not as slow as reusing a finger would be. It's not something I'd have to think about; this is already what I do whenever I have to type "rg" as part of a word.
I'm curious whether an ad-hoc approach is more or less efficient overall. Fingering customized per word clearly has the potential to optimize finger movement, but my error rate is relatively high - mostly timing-related - which might be exacerbated by an ad-hoc system because there are more (and more complex) unfamiliar transitions between words.
Anyway, I upvoted you for mentioning typing. It might seem trivial - well, if you use a lot of custom aliases, it is trivial - but if a command runs fast enough (and ag is already very fast on small source trees), the time spent typing its name can become a significant bottleneck. The author of Pijul, for instance, a version control system meant to compete with Git, seems not to recognize this... the command is 'pijul', which is essentially impossible to type on QWERTY without reusing a finger at least once.