[tex-live] Raspbian texlive

[simon at getthingsfixed.co.uk]

On Monday 10 September 2012 16:50:17 Boris Veytsman wrote:
>
> S> Is that all there was to it? Apart from about 8hrs runtime it wasn't any
> S> hassle to compile.
>
> Did you compile clisp/xindy and asy?  (not that I imagine someone
> would run xasy on this, but for completeness...)

No. I just did the bozo thing of "bash Build" with no params to see if 
anything interesting came out of the other end.

Wheezy has clisp as an package, so I installed that just now.

But presumably "bash Build --enable-xindy" is the right commandline and then 
wait for it to cook.

> S> If it is apparently that easy would I be making a foolish promise
> S> to supply the official armhf binaries for texlive?
>
> We did have requests for armhf binaries.  Arm7l binaries do run on
> raspbian, but LuaTeX seems to have problems since it uses floating
> point, which is implemented on armhf in a peculiar way.  

As reported earlier: the armel pdftex seems to work fine on Raspian. Probably 
cos it's all 32bit integer. But of course luatex can be more chilled and use 
64bit fp when it wants to. Which is fine if your fp is working, which in that 
case it isn't.

Tested luatex for numeracy and it's fine:
%%%%%%%%%%%%%%
--[[!
	\file
	\brief test some fp
	]]
--main
a = 1.1
io.write('a=' .. a .. '\n')
b = a * 1.1
io.write('b=' .. b .. '\n')
--eof
outputs:
a=1.1
b=1.21
%%%%%%%%%%%%%%

How critical is the exact build of Raspian going to be? Or is it all vanilla 
to a runtime binary?

> So a separate 
> build for armhf is in the plans - at least in my plans.  I even
> bought my own device, but it is still on the waiting list :(
>
> The final decision is up to Karl, but my opinion is yes, it would be
> very nice to have armhf binaries.

Mine, with xindy, should be cooked by tomorrow.


-- 
Get Things Fixed