From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <82c890d00706220037n6f64ee85pb30f8aedf9ed759e@mail.gmail.com>
Date: Fri, 22 Jun 2007 09:37:03 +0200
From: "Gabriel Diaz" <gabidiaz@gmail.com>
To: "Fans of the OS Plan 9 from Bell Labs" <9fans@cse.psu.edu>
Subject: Re: [9fans] printing via local attached printer with gs driver
In-Reply-To: <20070618180031.7874F1E8C4D@holo.morphisms.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
References: <82c890d00706181054s3ba18e59tdc4127bc232e7358@mail.gmail.com>
	<20070618180031.7874F1E8C4D@holo.morphisms.net>
Topicbox-Message-UUID: 84c04d60-ead2-11e9-9d60-3106f5b1d025

hello

I recompiled taking this in account, but the problem persists (the
src/gshtscr.c has a sightly different line in the distribution than in
the mkfile, so i changed it to the one of mkfile )

Also sources already have the make_other_poles function patched.

thanks

gabi


On 6/18/07, Russ Cox <rsc@swtch.com> wrote:
> Did you compile your own copy of ghostscript?
> This is a common problem if you compile directly
> from the gs sources; there are some Plan 9 specific
> changes that must be made.  See /sys/src/cmd/gs/mkfile:
>
> # If you get weird floating point errors, the culprit is usually the
> # halftone code, which converts double to uint, something 8c handles
> # incorrectly (treats as double to int).  Look in src/gshtscr.c for a line like
> #
> #       sample = (ht_sample_t)((value+1) * max_ht_sample);
> #
> # and change it to one of:
> #
> #       sample = (int)(vlong)((value+1) * max_ht_sample);
> #       sample = (ht_sample_t)(value * max_ht_sample) + max_ht_sample;
> #
> # depending on your preference.
> #
> # Also, recent versions of src/gxshade1.c cause the compiler to run out
> # of registers.  Brucee is looking into this.  In the meantime, use this
> # replacement:
> #
> #       private inline void
> #       make_other_poles(patch_curve_t curve[4])
> #       {
> #           int i, j;
> #
> #           for (i = 0; i < 4; i++) {
> #               j = (i + 1) % 4;
> #               curve[i].control[0].x = (curve[i].vertex.p.x * 2 + curve[j].vertex.p.x);
> #               curve[i].control[0].x /= 3;
> #               curve[i].control[0].y = (curve[i].vertex.p.y * 2 + curve[j].vertex.p.y);
> #               curve[i].control[0].y /= 3;
> #               curve[i].control[1].x = (curve[i].vertex.p.x + curve[j].vertex.p.x * 2);
> #               curve[i].control[1].y /= 3;
> #               curve[i].control[1].y = (curve[i].vertex.p.y + curve[j].vertex.p.y * 2);
> #               curve[i].control[1].y /= 3;
> #               curve[i].straight = true;
> #           }
> #       }
> #
> # (the original includes the /3 in the big expressions).
>
> The source in the distribution should be patched already.
> Also I thought I fixed 8c to do the right float -> uint
> conversion, but it would have been quite a while ago
> and I don't remember.
>
> Russ
>
>