Ihor Radchenko <[email protected]> writes: > Morgan Smith <[email protected]> writes: > >> In my last thread, I tried to convince you that we shouldn't be loading >> the loaddefs file and instead we should just use 'declare-function'. Ya >> that's a pretty stupid take. Doing that makes it way too easy to rely >> on functions that might not be loaded yet. >> >> Honestly I'm starting to think we should get rid of as many >> 'declare-function' calls as possible. The attached patch successfully >> removes 216 of them by loading the loaddefs file. > > I have a problem with this idea - it will make it *even easier* to rely > on functions that are autoloaded. You won't even have to add > declare-function! > > Even though I see instances of requiring loaddefs in Emacs code, I do > not think that it is a good practice from maintenance perspective. > It will hide circular dependencies even more.
Fair enough. A while ago I had a discussion with core Emacs devs about what should be autoloaded. The current documentation says only interactive functions should be but the devs said reasonable entry points to libraries can also be. I suppose use within an individual library should be quite limited... > I would really like to find some time to start merging my refactoring > branch, where circular dependencies are less problematic. Canceled. There is no easy fix. Only the hard proper fix. > However, but with the current rate of incoming reports, it is a bit hard > to find free time. Would you be willing to help me with making sure that > we always reply to the reports and followups to the eisting reports > within 1 month? The liaisons do help, but they do not cover all the > reports, unfortunately. I can try but no promises. Life and Guix sometimes get in the way :P My current workflow isn't amazing though. Bastien's GNAW package is very wonderful and helpful. I use gnus but it has some serious performance issues when dealing with this many emails. I should probably switch to mu4e or something but that's going to be a whole thing.
