NRK
I've experimented a bit around with this approach but it doesn't seem to be any faster than the hashtable. Using `ch
I don't think it's doable because the whole decoded buffer isn't kept in memory. https://github.com/vapier/ncompress/issues/27 has a decoder that basically stores each string in a flat array in their full...
> But my problem with this is that the worst case memory usage is unknown. I did a bit of math-ing today and TLDR is that the worst case is...
> I don't know if this issue is "theoretical" anymore Just to confirm whether your issue is the same as this or not, you can apply the following patch and...
What about `char num[STBSP__NUMSZ] __attribute__((aligned));` ? Does that work?
> I think this is what was meant (mind the "2"): Shouldn't make any difference. Without an "argument" aligned attribute aligns to ["maxium useful alignment"](https://gcc.gnu.org/onlinedocs/gcc-3.3/gcc/Type-Attributes.html) (which is likely `_Alignof(max_align_t)`). >...
> ffmpeg then internally removes the file:// prefix and does a open() call. Shouldn't this be fixed on ffmpeg's side? If it's accepting URIs then it should do it properly...
@kdsjZh Hi, could you please provide the output of `./w3m -version`, I'm not able to reproduce on my system. And the suspect code snippet is filled with various `#ifdefs` so...
Also since I'm not able to reproduce on my end, please also check if the following patch fixes the issue or not: ```diff diff --git a/etc.c b/etc.c index 805bfa0..bbad19c 100644...
> Github is telling me that you need to review this before the patch can be merged into main Huh, I think that's just some generic UI thing. I'm not...