Re: Euro currency support

On Wed, 17 Apr 2002 Dietmar.Buerkle@tsp-gis.de wrote:
>
> Ups, I forgot to mention:
> There is an ISO codepage iso8859-15 (aka Latin9) for western europe that is
> meant as a replacement for iso8859-1. SuSE-Linux supports this (for most
> fonts), debian too (at least there is an howto
>      http://www.debian.org/doc/manuals/debian-euro-support/)
> Recode 3.5 knows iso8859-15 so "the workaround" would work on linux too.

It appears that FreeBSD also supports ISO8859-15, which is good...

> All the CCSIDs I mentioned yesterday are Latin1 based (western eropean).
> But I found out there are at least 3 more CCSIDs with euro currency symbol:
> 1153 east europe, 1154 Russia, 1155 Turkey. Especially for eastern europe
> everything is much more complicated: Iso8859-16 is the replacement of
> Iso8859-2 with euro currency symbol, but they differ too much. I think in
> this case we either have to manually create a translation table or to wait
> for recode to support it.
>
> I have the following suggestion on creating transmaps.h:
>
> If the perl script transmaps in subroutine "recode" wants to write a maping
> say ibmxxxx_to_iso_8859_xx then we first should look for an file
> ibmxxxx_to_iso_8859_xx.map that is included into transmaps.h without ever
> calling recode. If the file is not found recode is called as usual.
>
> If someone needs an "complicated" mapping like ibm1153 to iso_8859-16 he
> could provide the (hand tuned) files iso_8859_16_to_ibm1153.map,
> windows_1250_to_ibm11531.map, ... BTW Marek would have an easy way to patch
> his transmaps.h too.
>
> Instead of writing "workaround" code in tn5250 to handle e.g. CCSID 1141
> almost like the pre euro CCSID 273 we (maybe me) could generate
> iso_8859_15_to_ibm1141.map, .. by an perl script (still based on 273). If
> one day recode knows the missing CCSIDs we just have to remove these map
> files.

The transmaps perl script is only run to generate a new transmaps.h.
Then, this is compiled into the emulator.

This means that if you created these special map files in a way that
requires the transmaps script to be run, only people building TN5250 from
source would be able to change the maps.

Instead, I recommend that if we have a map file, it's something that
can be read at runtime instead of at compile time!  Then, any TN5250 user
can change the maps, even make special maps to do special things!

>
> What do you think? (Hope you understood what I was trying to explain.)
>

I like the idea....  but I still think it would be even better if people
could change the maps without recompiling TN5250.

That's my 0.02 (insert euro symbol here)