[tex-k] Running (g)make check on FreeBSD: lib/t1lib, utils/tpic2pdftex

[Peter Breitenlohner]

On Thu, 10 Dec 2009, Nikola Le?i? wrote:

> 1. libs/t1lib
>
>   - additional -lm linking is necessary, for the test only; I am not
>     sure if this is an acceptable way to achieve this and if this is
>     a *BSD-specific issue;

Hi Nikola,

nice to know that I am not the only person doing 'make check'.

The need for libm seems to be *BSD-specific (or maybe it is Linux-specific
not to require libm for sqrt, floor, and similar math functions).  Please
tell me which functions were unresolved (probably sqrt, floor, ...), such
that I can add a suitable test in configure.  I wouldn't mind
unconditionally adding '-lm', except that I seem to remember that some
systems neither need nor have that library.

>   - BSD sed does not support "sed FILE -e '...'" syntax; the patched
>     version works for both sed and gsed.

Thanks for catching this GNU-ism, will be fixed soon in SVN.

> 2. utils/tpic2pdftex
>
> Here, example.exp does not match generated example.tex (beamerexample is
> ok). First I thought that it was a FreeBSD-specific problem, maybe with
> FreeBSD's pic(1), but it is not. FreeBSD's pic(1) and Linux pic produce
> the same tpic. So I needed this:
>
> Index: source/utils/tpic2pdftex/example.exp
> ===================================================================
> --- source/utils/tpic2pdftex/example.exp	(revision 16326)
> +++ source/utils/tpic2pdftex/example.exp	(working copy)
> @@ -402,8 +402,8 @@
> 143.496 -14.472 m
> 143.496 -14.472 l
> S
> -141.984 -11.232 m
> -141.984 -11.232 l
> +142.056 -11.232 m
> +142.056 -11.232 l
> S
> 139.824 -8.424 m
> 139.824 -8.424 l
> .....

I'd guess these differences are due to floating point rounding. Probably
best to only test if tpic2pdftex succeeds and ignore the differences.

Regards
Peter Breitenlohner <peb at mppmu.mpg.de>