[Xcb] more on the libGL port to XCB

Jeremy Kolb jkolb at brandeis.edu
Fri Apr 1 19:19:24 PST 2005


Jamey Sharp wrote:
> On Wed, Mar 30, 2005 at 01:51:29PM -0500, Jeremy Kolb wrote:
> 
>>Jeremy Kolb wrote:
>>
>>>Jeremy Kolb wrote:
>>>
>>>>um.... Can you check out GetBooleanv? It's on page 59 of the requests.
>>>>
>>>>Basically it says that if the length is one thing then this follows:
>>>>
>>>>1 BOOL params
>>>>15 unused
>>>>
>>>>else this follows:
>>>>
>>>>16 unused
>>>>n LISTofBOOL params
>>>><padding>
>>>>
>>>>I'm assuming this is the same for all Get*v replies.
>>>
>>>Ok, all these Get* functions are really weird.  The reply structures 
>>>change depening on if they succeed and what the length is.
>>
>>Any thoughts on this?
> 
> 
> Yes, but they're more of the unprintable variety.
> 
> This isn't very hard to encode, in simple form; it'll just be a bit
> tricky to use. Using the GetBooleanv example, the description should
> look something like:
> 
> BOOL param
> pad 15
> list of BOOL params

:(

> 
> It's then the caller's responsibility to figure out whether to use the
> 'param' field or the 'params' list.
> 
> We could extend the protocol description in various ways to make this
> simpler. In particular, it'd be nice if the generated GetBooleanvParams
> function would decide which part to return a pointer to based on the
> appropriate condition. It's worth considering, I think, but probably
> you'll want to adopt the approach that actually works right now. :-)
>

:(  Gross.


Jeremy


More information about the xcb mailing list