# [tex-k] [reinhard.kotucha at web.de: Re: [texhax] can't control page size - why? (the real problem)]

Tom Rokicki rokicki at gmail.com
Wed Apr 26 00:19:06 CEST 2006

That's fine.  We can also make it just a tiny bit smarter: instead of
finding the first match within the accuracy bound, it finds the best
match within the accuracy bound.  This would be a straightforward
change to make.

On 4/25/06, Karl Berry <karl at freefriends.org> wrote:
> Perhaps it would make sense to implement this (5bp) as dvips' criteria for
> "close enough"?  Tomas, Akim?
>
> (Thanks Reinhard!)
>
> k
>
> Date: Tue, 25 Apr 2006 22:35:12 +0200
> From: Reinhard Kotucha <reinhard.kotucha at web.de>
> To: karl at freefriends.org (Karl Berry)
> Cc: toms at ncifcrf.gov, texhax at tug.org
> Subject: Re: [texhax] can't control page size - why? (the real problem)
>
> >>>>> "Karl" == Karl Berry <karl at freefriends.org> writes:
>
>   > By sheer coincidence (?), another fellow posted to Debian and
>   > thence to tex-k with the problem that dvips was "rounding" paper
>   > sizes to the closest "standard" size.  Perhaps that is happening
>   > here.  Just a wild guess ...
>
> Maybe it's best to follow the rules for the /PageSize key in the
> setpagedevice dictionary.  A4 paper will be chosen if you say
>
>    <</PageSize [595 842]>> setpagedevice
>
> and the values of width and height do not differ by more than
> $\pm$ 5 bp from a particular standard (known) paper size.
>
> I do not know where the problem occurs, PostScript itself needs only
> numbers and no rounding is needed.  Strings like (A4) are needed for
> DSCs, but in this case it makes much sense if a particular paper size
> is recognized as A4 if, and only if, it is recognized as A4 by
> setpagedevice as well.
>