making X server a COM object..

[Harold L Hunt II]

KH,

What is am saying is that COM, by itself, is something you have to sit 
down with for a few months before you even realize if what you are 
suggesting is possible.  No amount of pointers from other people are 
going to help.

Similarly, X is something that you have to sit down with for a few 
months before you can determine if such things are possible.


We can talk about threading, ATL, etc... but it won't matter until you 
are presenting a coherent framework (which would take about 20 pages to 
describe).  At that point, feel free to ask, "what did I forget?". 
Understand though that asking us "will this work" is too open-ended of a 
question.  We just can't spend time on your question until you 
demonstrate that you have spent a lot of time with it already.

Harold

Chan Kar Heng wrote:
> thank you for the courtesy.. :)
> .. and i would likely (85%) agree with you that i do not have a
> firm grasp on the scope... probably more.. probably less..
> but if i had a firm grasp, i probably wouldn't need to ask around
> the gurus here right? :)
> 
> what i do hope is to be able to get some pointers directly at
> potential problems... and why it might be impossible to solve...
> rather than spending the time to research each one of them
> (cygwin, cygwin xfree, COM, etc) in detail to look for potential
> problems...
> if the issue might be a lengthy one... i suppose some urls or
> general description of the issue would suffice.
> 
> some potential problems that i can foresee from what i picked up
> so far:
> 
> - different and likely conflicting runtimes... atlcrt, msvcrt, cygwin1.
> if i'm not mistaken, in this case, it's possible to have the cake
> and still eat it... but it's very very tricky...
> - threading...
> - COM on cygwin (even COM on mingw is difficult enough.. and
> one can't use C++ due to different name mangling technique).
> 
> do correct me in areas where i might be wrong...
> 
> havn't been following on cygwin xfree86... so don't know much
> bout it...
> 
> comments?
> 
> 
> rgds,
> 
> kh
> 
> 
> 
> 
>>Not to be rude... but I don't think you have a firm grasp on the scope of what you are talking about.
>>
>>Harold
>>
>>Chan Kar Heng wrote:
>>
>>>hmmm... i'd certainly avoid porting XFree86 to use win32 instead of cygwin...
>>>i have a feeling it'd be a *lot* of effort...
>>>i'm thinking.. if it might work if all the .dll files and XWin.exe required were
>>>compiled to .o or .a files.. then using a COM object, statically link in all
>>>those .o or .a files... (it would be a pretty huge COM obj though, but it's
>>>a start at least?)...
>>>
>>>At 2003/04/01 07:58, you wrote:
>>>
>>>
>>>>Harold,
>>>>
>>>>At 15:50 2003-03-31, you wrote:
>>>>
>>>>
>>>>>KH,
>>>>>
>>>>>The scope is probably beyond the scope of this mailing list.
>>>>>
>>>>>I think you would be better off working first on a version of Cygwin/XFree86 that compiled without Cygwin... then, and only then, could you even begin to worry about wrapping XFree86 with a COM interface.
>>>>
>>>>Cygwin /XFree86 without Cygwin is Win32/XFree86, or some such, is it not?
>>>>
>>>>It always seemed to me that the target specifier ought to come after the main program name: XFree86/Cygwin, XFree86/Win32, etc.
>>>>
>>>>
>>>>
>>>>
>>>>>Harold
>>>>
>>>>Randall Schulz 
>>>
> 
>