[PATCH v2 1/6] kernelbase/console, programs/conhost: support CONSOLE_READCONSOLE_CONTROL in ReadConsoleW

[Jacek Caban]

On 2/27/22 15:02, Eric Pouech wrote:
> Le 25/02/2022 à 20:10, Jacek Caban a écrit :
>> Hi Eric,
>>
>> The series looks mostly good for me.
>>
>> On 2/23/22 18:05, Eric Pouech wrote:
>>> +    if (initial_len && edit_line_grow( console, initial_len + 1 ))
>>> +    {
>>> +        unsigned offset = edit_line_string_width( initial, 
>>> initial_len );
>>> +        if (offset > ctx->home_x)
>>> +        {
>>> +            int deltay;
>>> +            offset -= ctx->home_x;
>>> +            deltay = offset / console->active->width;
>>> +            ctx->home_y = (deltay >= ctx->home_y) ? ctx->home_y - 
>>> deltay : 0;
>>
>>
>> Isn't the logic reverted here?
> yes
>>
>>
>>> +            ctx->home_x = console->active->width - 1 - (offset % 
>>> console->active->width);
>>
>>
>> I wonder if we should use use 0 for home_x in case of deltay >= home_y.
>
> actually, this was more to protect the rest of the code which isn't 
> prepared for negative coordinates
>
> IIRC I tested this UC and native supports this
>
> changing the code for handling such case is not immediate, but I 
> thought that the ratio real value vs effort was near 0 and not worth 
> for now going for it
>
> so we could either add a FIXME (and leave the 0) or add a FIXME and 
> return an error 


I commented mostly based on the line above (which seems more important). 
If it's closer to what we will eventually want (if we support negative 
home_y someday), that's fine with me, as long as we don't crash.


Thanks,

Jacek