Re: ELPA policy

[Dmitry Gutov]

On 11/10/2015 10:34 PM, Eli Zaretskii wrote:

> Grep doesn't scale well to large projects, IME.  You get too many
> false positives.

We can also check the "inside string or comment" status, as well as use 
the Emacs syntax tables to make sure that the match begins and ends at 
symbol boundaries appropriate for the file's language. Everything 
necessary for that is already written (and you can try it out by calling 
project-find-regexp and using something like \_<tool_bar_items\_> as the 
regexp). I don't think that id-utils does anything more to weed out 
false positives.

Grep is still probably going to be slower than at least some of the 
tools in question. Could you test, on a large project of your choice?

> Outdated databases are easy to avoid with the likes of cron jobs.
> Yes, that's hand-holding, but when you have to quickly find stuff in a
> project with 3 million lines of code and thousands of classes, there
> really is no other alternative.

Yes. But the xref commands should be easy to use. Even if the above is 
not rocket science, the user would still have to know what they need to 
do to get up-to-date results. (Believe it or not, I haven't created a 
single cron job for code writing purposes in my life, and I don't know 
its syntax for the time intervals. I'm likely not alone in that.)