endif without if in context.vim file

[RanjitV17]

In file autoload/context/context.vim, in s:limit function, there is an extra endif towards the end without if.

[wellle]

Thank you! Do you want to open a PR to remove it?

[RanjitV17]

You please go ahead and fix this as part of your next planned release. Thank you so much for your great work. I use targets too.

May I know the reason why you wanted to skip the pre-processor directives in C code? IMO they add a lot of value to the context.

[wellle]

Thanks, will do 👍

May I know the reason why you wanted to skip the pre-processor directives in C code? IMO they add a lot of value to the context.

Sure. The context currently works based on indentation. Because the pre-processor directives often have zero indentation that means that the context will effectively have no more context lines above. A bit hard to explain, here's an example. Consider this code: https://gist.github.com/wellle/981686e250e6bc1851b78fbcc6dc6b3a#file-v-c-L126

Currently the context of the highlighted line is

find_some_match(argvars, rettv, type) ··· {
    if (regmatch.regprog != NULL) {
        for (;;) {
            if (l != NULL) { ··· }
            else {

If I update the skip regex to effectively allow the # lines, then the context looks like this:

#ifdef FEAT_MBYTE

It might be possible to come up with more complex logic to allow both pre-processor directives and retain the full context from above, but that would be a lot of extra work. For now I had just decided to exclude those lines, as in my example above the result seems more useful. Hope that makes sense.

[RanjitV17]

Got it. I updated the skip regex to allow the # and I see the same behavior as you mentioned. I wasnt aware that the context update is heavily relying upon the indentation. I was thinking more in terms of C code only not in a generic way.
If this was made to work only with C, may be we can rely only on keywords + braces etc regex and not on indentation. Am I right here?

[wellle]

Fixed in #76. Thanks again!

And regarding your last comment, yes you're right, the context itself could definitely be smarter if it were implemented in a language specific way. But so far I opted for a general purpose approach and I'm hoping that it's going to be good enough 🤞