Re: [Emutos-devel] Splitting the 512K ROMs

[Roger B. <an...@xp...>]

On 10 Feb 2021 at 15:29, Thorsten Otto wrote:

> On Mittwoch, 10. Februar 2021 03:23:29 CET Roger Burrows wrote:
> 
> I think most has already been said. The only situation were a multi-language
> TOS might be useful is, when you not only supply EmuTOS, but a complete 
> "distribution", like eg. the minipack. In that case, using a multi-language 
> capable EmuTOS that works for everyone is very handy.
> 
> OTOH, some time ago i already started to rework the multi-language feature 
> (not yet completed). I came to the conclusion that is should be possible to 
> get rid of all the hash & lookup tables. All that remains is a lookup table 
> per language (about 200 bytes for each language), plus the translated texts 
> (about 8-9K per language currently). That should free some considerable room
> in the 512k ROMs, and also get rid of the runtime-overhead of looking up the
> strings.
> 
Yes, the current method is pretty ugly, but presumably the hash tables were 
chosen for speed.  Without testing, I'm not sure how important that is versus 
doing a binary search.  But in any case, considering there are 300+ text 
strings for each language, getting the lookup info into 200 bytes sounds pretty 
impressive.  I hope you can get it done soon!

Roger
P.S. Even if you manage to get rid of the run-time lookup for most strings, 
there are still explicit runtime references to gettext(), so I'm curious how 
you're going to fix that.