Re: [hackers] [sbase] [PATCH 00/10] IO improvements and some bug fixes

From: Michael Forney <mforney_AT_mforney.org>
Date: Mon, 5 Dec 2016 01:22:57 -0800

On Mon, Dec 5, 2016 at 12:51 AM, Michael Forney <mforney_AT_mforney.org> wrote:
> On Sun, Dec 4, 2016 at 11:27 PM, Laslo Hunhold <dev_AT_frign.de> wrote:
>> fgetc/fputc really can slow down your program, which I noticed when I
>> wrote farbfeld-filters. Working on a buffer-basis rather than a
>> char-by-char-basis really speeds up things a lot (for me it was around
>> 50%-70%).
>
> Internally, fgetc/fputc are still using the buffers behind the FILE
> structures, so the only overhead I can imagine is due to the repeated
> function calls. The previous tail -f code used getline/fputs. We could
> fix the NUL byte issue with fwrite instead of fputs, but I'd like to
> make sure that it actually makes any noticeable difference before
> trying this.

So I think you are correct in that this is actually horribly slow. I
think tail should probably be managing its own buffer. I might take a
look at what other tail implementations do and play around with this.
For now, I think getline/fwrite is the only thing to do since fgets
doesn't return the string length, and fputs will stop at the first
NUL.
Received on Mon Dec 05 2016 - 10:22:57 CET