Re: guile-2.0 and debian

[Thomas Morley]

2016-11-21 13:09 GMT+01:00 Antonio Ospite <address@hidden>:
> On Mon, 21 Nov 2016 10:59:17 +0100
> Thomas Morley <address@hidden> wrote:
>
>> 2016-11-21 10:49 GMT+01:00 Antonio Ospite <address@hidden>:
>>
>> > Can I look at your ./56/lily-44b4b5eb.eps file?
>>
>> Here you are.
>>
>
> The problem is that there are still some numbers formatted with commas
> as decimal points when producing EPS: it shows in the standalone
> command line too if you use '-dbackend=eps'.
>
> The following change fixes it:
>
> ---------------------------------------------------------------------
> diff --git a/ly/init.ly b/ly/init.ly
> index 58b2fac..8fdc6e2 100644
> --- a/ly/init.ly
> +++ b/ly/init.ly
> @@ -9,7 +9,8 @@
>  #(if (guile-v2)
>    (begin
>     (use-modules (ice-9 curried-definitions))
> -   (setlocale LC_ALL "")))
> +   (setlocale LC_ALL "")
> +   (setlocale LC_NUMERIC "C")))
>
>  #(session-initialize
>    (lambda ()
> ---------------------------------------------------------------------
>
> And this will also replace my previous ugly attempt of messing around
> with LC_NUMERIC in lily/general-scheme.cc.


I reverted locally "Print floating point variables using a period as
the decimal separator" and applied your code above.
I got a successful "make LANGS='' doc"

That's great!

Next would be a full "make doc" or at least with the Japanese docs to
proof non latin.
And I'd like to do a regtest-comparison, between builds with guile
1.8.8 and 2.0.13

Although, I currently don't know if this is possible at all with our
"make test-baseline"/"make check"-procedures.
Will try tomorrow.

If all this only would not last ages on my weak laptop...


Thanks,
  Harm