Hi all,
I am packaging [1] sunpinyin. Sunpinyin is a Chinese input method engine.
As you may know, this source package builds two binary packages for two
input method platform (SCIM and IIIMF) respectively. Both of these binary
packages contain a copy of the lexicon and its index data files. The total
size of the data files is about 19M in each binary package. And what makes
things more complicate is that there are actually two copies of data in
source package. One is in big-endian format, the other is in small-endian
format. So the source package is around 38M.
As the size of data files in the package is considerable, I think it is
necessary to split the data out into separate package(s) [2]. So far, the
best approach in my mind is to create two other data packages named
sunpinyin-le-data and sunpinyin-ge-data, and make the binary packages depend
on the data package by its endianness accordingly.
Is this solution feasible, or is there any better way to solve this problem?
And moreover, I am afraid that sunpinyin-{le,ge}-data are not architecture
neutral. So is it okay to make these two packages Arch:all?
At last, Is there any way to specify the package dependency by architecture?
Thanks in advance.
--
[1] Bug#478811: ITP: sunpinyin -- An input method engine based on SLM
[2]
http://www.debian.org/doc/developers-reference/ch-best-pkging-practices#s-bpp-archindepdata
--
Regards
Kov Chai
