Jan. 26, 2026
11:57 p.m.
On Mon Jan 26 23:57:18 2026 +0000, eric pouech wrote:
> thanks for taking care of this evolution
> just a cursory look (far from being a review):
> * style: please don't use C++ comments, nor case switch ranges (they are
> not portable enough)
> * style: try to stick to the current style of modified file
> * size_t shall be printed with %Iu
> * escape sequences should only be interpreted when console output mode
> has the vt seq bit set
> * some non regression tests could be useful (no precise idea on what
> they should cover)
> * reusing SET_CONSOLE_OUTPUT_INFO_ATTR is a tempting idea; did you check
> that how is done on native? (it's a matter of knowning if the "current"
> color from an ANSI seq is also the default screen buffer color; or when
> two programs are attached to the same conhost, they share the "current"
> ANSI color)
> * MR should be tested for both wineconsole and unix console
- style: please don't use C++ comments, nor case switch ranges (they are not portable enough)
- switch case ranges removed in local branch
- style: try to stick to the current style of modified file
- style and comments adjusted in local branch
- size_t shall be printed with %Iu
- i had a conflict for size_t on my local machine vs. build pipeline build32
```
%lu for size_t?!
on my 64bit system
size_t {aka long long unsigned int}
VS
pipeline build32
size_t {aka unsigned int}
```
- escape sequences should only be interpreted when console output mode has the vt seq bit set
- Is ENABLE_VIRTUAL_TERMINAL_PROCESSING the correct bit? It is not checked for other sequences like \t or \n
- some non regression tests could be useful (no precise idea on what they should cover)
- if you have some suggestions or hints how this should look like, i will try to implement some test
- reusing SET_CONSOLE_OUTPUT_INFO_ATTR is a tempting idea; did you check that how is done on native? (it's a matter of knowning if the "current" color from an ANSI seq is also the default screen buffer color;
- the basic SGR sequences use ANSI color, but the 38 and 48 commands can use RGB values too, therefore my simple approach to convert RGB to ANSI color. Midnight Commander uses this RGB values for setting color, but seems to stick to ANSI colors, but i am pretty sure that MC allows to modify its color scheme
- or when two programs are attached to the same conhost, they share the "current" ANSI color)
- what do you mean by share current ANSI color? each screenbuffer stores its color per character/data, this should not conflict between multiple consoles
- MR should be tested for both wineconsole and unix console
- how can i test unix console? do i have to read the code to figure it out or is there a wiki/howto?
**In general:**
Should i push additional commits to the branch or should i rebase and fix the already existing commits?
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/9973#note_128075