view DOCS/tech/subcp.txt @ 33644:c15dabfa2380

Do a proper calculation of free RAM to be used as V4L buffers. The code uses sysinfo to query the available RAM, however it used ancient form available is some early development 2.3.x kernels. Newer form reports the size in memory units (usually same as page size), as result the code would fallback to 2 buffers even on multi GB system. The commit does: Improve the check in configure to ensure that we do use sysinfo struct with present mem_unit. Use free ram instead of total ram (to avoid swapping). Tweak memory constants and simplify code.

Ascii Subtitle / Font CODEPAGEs
===============================

The subtitle encoding issue seems a bit confusing, so I'll try to
summarize it here.

There are 2 approaches:

1. (preferred) You can generate Unicode subtitles with:
        subfont --unicode <signle-byte encoding known by iconv> ...
or
        subfont --unicode <path to custom encoding file> ...
        (this custom encoding file could list all iso-8859-* characters to create
single font file for common encodings)

and then run mplayer this way (-subcp and -utf8 expect Unicode font!):
        mplayer -subcp <any encoding known by iconv> ...
or
        mplayer -utf8 ...

2. (current) Generate subtitles for some specific encoding with:
        subfont <signle-byte encoding known by iconv> ...
or
        subfont <path to custom signle-byte or EUC encoding file> ...

and then run mplayer without any encoding options for signle-byte
encodings, or with -unicode option for EUC (and the like) encodings
(which is only partially implemented in mplayer).

AFAIK, CJK encodings: EUC-*, BIG5 and GB2312 work more or less this way:
- 0x8e (SINGLE-SHIFT TWO, SS2) begins a 2-byte character,
- 0x8f (SINGLE-SHIFT THREE, SS3) begins a 3-byte character,
- 0xa0-0xff begin 2-byte characters,
- other characters are single-byte.


I tested charmap2enc script only with /usr/share/i18n/charmaps/EUC-KR.gz
(on RedHat). It wasn't intended to be perfect.


--
Artur Zaprzala