Differences From Artifact [b6e40ac9ff]:
- File
src/OFStdIOStream_Win32Console.m
— part of check-in
[44f45c2e35]
at
2017-01-09 17:36:36
on branch trunk
— Update copyright
Forgot to add 2017, even though I already did quite some changes in
2017. (user: js, size: 8026) [annotate] [blame] [check-ins using]
To Artifact [7220727656]:
- File src/OFStdIOStream_Win32Console.m — part of check-in [5a31a537a1] at 2017-02-04 17:45:33 on branch trunk — Fix typos in comments and strings found by aspell (user: js, size: 8027) [annotate] [blame] [check-ins using]
| ︙ | ︙ | |||
23 24 25 26 27 28 29 | * For example, on Windows XP, when using Windows XP's console, changing the * codepage to UTF-8 mostly breaks write() and completely breaks read(): * write() suddenly returns the number of characters - instead of bytes - * written and read() just returns 0 as soon as a Unicode character is being * read. * * Therefore, instead of just using the UTF-8 codepage, this captures all reads | | | 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 |
* For example, on Windows XP, when using Windows XP's console, changing the
* codepage to UTF-8 mostly breaks write() and completely breaks read():
* write() suddenly returns the number of characters - instead of bytes -
* written and read() just returns 0 as soon as a Unicode character is being
* read.
*
* Therefore, instead of just using the UTF-8 codepage, this captures all reads
* and writes to of_std{in,out,err} on the low level, interprets the buffer as
* UTF-8 and converts to / from UTF-16 to use ReadConsoleW() / WriteConsoleW().
* Doing so is safe, as the console only supports text anyway and thus it does
* not matter if binary gets garbled by the conversion (e.g. because invalid
* UTF-8 gets converted to U+FFFD).
*
* In order to not do this when redirecting input / output to a file (as the
* file would then be read / written in the wrong encoding and break reading /
|
| ︙ | ︙ |