[BUG] Lister: Automatic detection of OEM charset

[icfu]

3 out of 500 of arbitrary OEM files I have tested were reported as ANSI, that's a 99,94% accurary rate. Impressive!

Icfu

[xoft]

How is this bug / feature doing? Is it at least considered? It has been quite bothersome for some time now.

[ghisler(Author)]

Unfortunately I haven't yet found a better detection method for plain texts with just a few accented characters, so I have left it as it is for now.

[xoft]

How so? Alextp has provided a code with quite an impressive accuracy rate.

[ghisler(Author)]

Alextp's code is good for finding pseudo-graphics characters only. It cannot be used t distinguish regular texts written with a DOS program from regular Windows texts...

[xoft]

I believe it is quite safe to assume that pseudo-graphics characters are the only thing that OEM mode will be used for nowadays - ASCII art. I can't really think of any DOS program that stores text files and is still widely used today.

[icfu]

So, because the code cannot detect ALL OEM files, you prefer to keep your own code which cannot detect a SINGLE OEM file correctly.

Not really convincing, especially regarding that plain texts written in DOS editors are way less frequent nowadays than files with pseudo graphics...

Icfu

[ghisler(Author)]

No, I just fear the many support mails from users where it _no longer_ works...

[xoft]

Easy then - keep the old algorithm but make it available only through settings, with default setting being the new algorithm.