From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mimir.eigenstate.org ([206.124.132.107]) by ewsd; Wed May 27 18:43:59 EDT 2020
Received: from abbatoir.fios-router.home (pool-162-83-132-245.nycmny.fios.verizon.net [162.83.132.245])
	by mimir.eigenstate.org (OpenSMTPD) with ESMTPSA id c38a063d (TLSv1.2:ECDHE-RSA-AES256-SHA:256:NO);
	Wed, 27 May 2020 15:43:35 -0700 (PDT)
Message-ID: <5505AF6CCE9AC7B6543F256861B0198F@eigenstate.org>
To: lucas.francesco93@gmail.com, 9front@9front.org
Subject: Re: [9front] zlib update
Date: Wed, 27 May 2020 15:43:33 -0700
From: ori@eigenstate.org
In-Reply-To: <CAF=5iUVsKSgWseEChs-X-7JenBU7URYgYi3AvKh+znM_XH6vQw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
List-ID: <9front.9front.org>
List-Help: <http://lists.9front.org>
X-Glyph: ➈
X-Bullshit: compliant virtualized polling realtime frontend 

> Today people asked me to resubmit the zlib patches that I had sent
> months ago due to netsurf/nsfb. I actually had things working and
> clean on my system since back then, but I was confused between a
> decision of sending the build "half-broken" but with the final result
> working, or triggering a flag up that I didn't know if it should or
> not, after looking up, it seems the cleaner route is just to do the
> latter.
> 
> the newer zlib uses lseek/open/write, functions specified in
> <unistd.h>, but it actually doesn't include the header when building
> some of the .$O, what happened it was just that our linker would spot
> the functions even after the compiler warning due to the lax
> prototypeless functions in ANSI C std and build it fine. (the -B flag
> on 2c would do it)
> 
> the option I'm going here is just to make it include <unistd.h>, that
> requires anything that uses ape/zlib to define _POSIX_SOURCE

To prevent someone else from making the same error, we can do something
like

#ifndef _POSIX_SOURCE
#error zlib requires _POSIX_SOURCE to be defined
#endif

in zlib.h

> all tests I've done points out it is working fine, page works, the zlib
> tests passes and netsurf png decoding work as always with it.

> +++ b/sys/src/cmd/gs/zlib/Makefile.in	Wed May 27 17:39:37 2020 -0300
> @@ -1,11 +1,11 @@
> +++ b/sys/src/cmd/gs/zlib/configure	Wed May 27 17:39:37 2020 -0300
> @@ -1,443 +1,921 @@

We can probably get rid of these files -- they're not useful.

Will look a bit deeper later, but other than that, the patch looks
big and hard to review to me, since it's such a big hump in versions.

It's also widely used upstream code, so.. *shrug*, it's probably ok.