'Re: [PATCH 09/14] dlls/gdi32: use correct integral type'

[prev in list] [next in list] [prev in thread] [next in thread] 

List:       wine-devel
Subject:    Re: [PATCH 09/14] dlls/gdi32: use correct integral type
From:       Jacek Caban <jacek () codeweavers ! com>
Date:       2022-01-31 16:34:03
Message-ID: b6df555c-4a0f-d037-802b-e3ab5824785e () codeweavers ! com
[Download RAW message or body]

On 1/31/22 14:44, Eric Pouech wrote:
> Le 31/01/2022 Ă  14:24, Nikolay Sivov a ĂŠcrit  :
>> Maybe we could adjust NtGdi prototypes instead. I see a number of 
>> casts like that in existing code already, that would be unnecessary 
>> if DWORD was used for the prototype.
>>
> I'm in for avoiding the casts !
>
> but, these APIs are not documented on MS web servers
>
> and I don't have a ddk handy to check the prototypes
>
> reactos and drmemory both define ntgdipolypolydraw with a PULONG 
> instead, which could be a hint, not an absolute confirmation though
>

I think that, since it's not documented anyway, we may pick the type 
that works best, as long as it's binary compatible. Note that some 
callers already have casts, because it's sometimes used on INT*, 
sometimes on DWORD*, depending on the function. We could avoid new casts 
by simply using DWORD or ULONG for NtGdiPolyPolyDraw.


Thanks,

Jacek

[Attachment #3 (text/html)]

<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <div class="moz-cite-prefix">On 1/31/22 14:44, Eric Pouech wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:63b0906d-c2dd-20d6-a25b-1b62191edd21@orange.fr">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <div class="moz-cite-prefix">Le 31/01/2022 Ă  14:24, Nikolay Sivov
        a ĂŠcrit  :<br>
      </div>
      <blockquote type="cite"
        cite="mid:2c8cec53-f045-4767-debb-2c412058d8f6@codeweavers.com">Maybe
        we could adjust NtGdi prototypes instead. I see a number of
        casts like that in existing code already, that would be
        unnecessary if DWORD was used for the prototype. <br>
        <br>
      </blockquote>
      <p><font face="Helvetica, Arial, sans-serif">I'm in for avoiding
          the casts !<br>
        </font></p>
      <p><font face="Helvetica, Arial, sans-serif">but, these APIs are
          not documented on MS web servers</font></p>
      <p><font face="Helvetica, Arial, sans-serif">and I don't have a
          ddk handy to check the prototypes</font></p>
      <p><font face="Helvetica, Arial, sans-serif">reactos and drmemory
          both define ntgdipolypolydraw with a PULONG instead, which
          could be a hint, not an absolute confirmation though</font></p>
    </blockquote>
    <p><br>
    </p>
    <p><font face="Helvetica, Arial, sans-serif">I think that, since
        it's not documented anyway, we may pick the type that works
        best, as long as it's binary compatible. Note that some callers
        already have casts, because it's sometimes used on INT*,
        sometimes on DWORD*, depending on the function. We could avoid
        new casts by simply using DWORD or ULONG for </font>NtGdiPolyPolyDraw.</p>
    <p><br>
    </p>
    <p>Thanks,</p>
    <p>Jacek<br>
    </p>
  </body>
</html>


[prev in list] [next in list] [prev in thread] [next in thread]
Configure | About | News | Add a list | Sponsored by KoreLogic