Alexey Dokuchaev

> Well, our code only returns `-I/usr/local/include/gphoto2` Okay, but if your code, and other consumers' code I've seen, would typically `#include `, wouldn't it make more sense to return `-I/usr/local/include`...

> for system includes the `/usr/include` would also already cover that. That's true for GNU/Linux, but some of the better operating systems install third-party software under different prefix, e.g. `/usr/local`...

Just have [added it](https://www.freshports.org/editors/turbo/) to the FreeBSD Ports Collection.

Also fine, I guess. Personally I prefer to avoid enumerating systems where something is supported, because e.g. tomorrow OpenBSD or Haiku hacker comes, and then you have to extend that...

The same bug had just been [fixed](https://cgit.freebsd.org/ports/commit/?id=735a7f11b1c53e58d7ab011b274724a84a8e5e15) in the FreeBSD port (more than two years after it was reported here).

I believe commit d17159f fixes this once and and for all, and could probably explain why commit 3aa91ea was sufficient for Pygments v2.9, but not for v2.10 and above.

Given that the latest commit a0d3fdf fixes buffer overflow, perhaps just tag it as version 0.4.1 or 0.5.0? This would greatly help downstream packagers.

Could be related, not sure (feel free to move to its own issue). There seems to be some race condition when pressing "down" key shortly after seeing feed contents: ```...

It can be triggered with GCC/binutils as well by putting `-Wl,--no-undefined-version` on `CFLAGS`. Yes, it would be nice to have this fixed.

_ftime(3)_ is historical interface and is usually obsoleted by _gettimeofday(2)_ on modern systems, maybe OpenBSD removed it completely. Consider the following pseudo-patch (comments and typecasts removed for brevity): ```patch ---...