I dunno, its been like that since before my time (eg that line in 0.18 is the same).

The purpose of the parsing is to identify lines in the compiler output containing filename and line number that the user might want to go to so they can investigate and fix the error. That is used to make the message line clickable to go to that code line. Its a user convenience.

Like I said above I believe the idea is to make as many lines as sensibly possible be clickable (ie file & line are detected no matter by which method) to allow the user to go to the any of those lines. Since many compilers spit out lots of lines for each error and the fix may be on any one of those lines. Lines that don't have file and line detected cannot be clickable because there is no indication of where to go, but all lines that have an identifiable filename and line number should be clickable even if they pertain to the same error. And as the extra lines often have subtly varying formats the ability to simply add detection for those on top of the standard detection is useful.

I wouldn't like to see any lines that are currently clickable be made un-clickable, that would be a significant reduction in functionality, and all the other IDEs that I have recently used allow any line that has a recognised file and line in it to be clicked to go to that line, no matter what the message was.

Agreed in the current circumstances next/previous error action actually goes to next/previous line that is clickable, since in compilers for simple languages or older compilers thats usually the next error, but for the obnoxiously verbose modern C/C++ compilers its not the case. But that would be harder to fix.

The only case I can see to not fallback if a regex fails is if the fallback is producing a lot of cases where it detects a filename and line number when its not one. That would be better disabled with a checkbox per filetype (even better per regex) rather than blowing away the functionality for all filetypes hard coded. Just because you don't want it shouldn't mean everyone loses it.

Or the other option is to totally remove the fallback code and provide an equivalent default regex in every filetype, PRs are welcome.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or unsubscribe.