Add multibyte / locale and ignore support #49
Conversation
|
This is gonna take a bit longer than expected, as (In addition to multibyte, another way column counting may break is... control characters in linebuf. editline now correctly show them in caret-escape form, so despite them being one byte they take 2 columns.) The mblen thing, on the other hand, might be a bit easier to deal with now. I figured instead of the 100% correct left-to-right approach (parse the whole string), I can go right to left without breaking too many non-self-synchonizing legacy encodings. |
| [East Asian ambiguous]: http://www.unicode.org/reports/tr11/tr11-38.html | ||
|
|
||
| The input handling is not yet multibyte-aware. Specifically, the presence of | ||
| non-self-synchonizing encodings requires always segmenting the whole string |
There was a problem hiding this comment.
s/synchonizing/synchronizing/ maybe?
| @@ -125,6 +125,9 @@ extern void rl_callback_handler_install (const char *prompt, rl_vcpfunc_t *lhand | |||
| extern void rl_callback_read_char (void); | |||
| extern void rl_callback_handler_remove (void); | |||
|
|
|||
| #define RL_PROMPT_START_IGNORE '\1' | |||
| #define RL_PROMPT_END_IGNORE '\2' | |||
|
|
|||
| @@ -61,6 +61,9 @@ AC_ARG_ENABLE(terminal-bell, | |||
| AC_ARG_ENABLE(termcap, | |||
| AS_HELP_STRING([--enable-termcap], [Use termcap library to query terminal size.])) | |||
|
|
|||
| AC_ARG_ENABLE(locale, | |||
There was a problem hiding this comment.
I think I'd prefer for this to be enabled by default, since that's what ppl expect, and detect capability and fall-back to non-utf-8 capable.
|
Why hasn't this been merged |
|
@mcfriend99 because it's unfinished? Op clearly states "This PR, when completed ..." and "This is gonna take a bit longer than expected ...", not to mention the unaddressed review comments of mine that are yet to be resolved. |
This PR, when completed, shall close #37 and close #48. The handling for multibyte in the prompt is done, but not for the input.