Re: [PATCH] Include from .

2007-06-08 Thread Jan Engelhardt
On Jun 8 2007 11:42, Robert P. J. Day wrote: >> > >> > um ... what does that mean? consider this snippet from >> > : That's right. Each .h file should be "compilable" on their own. >> >> Which compile combination shows the sequence you are reporting ? > >sorry, i don't understand the question.

Re: [PATCH] Include from .

2007-06-08 Thread Alan Cox
> ok, i'll defer to your experience. but can we at least apply my patch > to protect console_struct.h from multiple inclusion? that one really > *is* annoying. I've not seen that one but send it - multiple inclusion is a different problem altogether - To unsubscribe from this list: send the

Re: [PATCH] Include from .

2007-06-08 Thread Robert P. J. Day
On Fri, 8 Jun 2007, Alan Cox wrote: > > if you don't agree with that patch, fine. but suggesting that it's > > going to result in "a mass of unresolvable dependancy loops" is sheer > > laziness. > > No its "sheer experience", frequently painful. The tty headers and > the scheduler headers are

Re: [PATCH] Include from .

2007-06-08 Thread Alan Cox
> if you don't agree with that patch, fine. but suggesting that it's > going to result in "a mass of unresolvable dependancy loops" is sheer > laziness. No its "sheer experience", frequently painful. The tty headers and the scheduler headers are particularly problematic for getting tangled.

Re: [PATCH] Include from .

2007-06-08 Thread Robert P. J. Day
On Fri, 8 Jun 2007, Alan Cox wrote: > > if a header file needs a definition/declaration that exists in another > > header file, it should include that other header file. any other > > approach is wrong. > > And if you try this in the real world for a project as large as the > kernel you will die

Re: [PATCH] Include from .

2007-06-08 Thread Alan Cox
> if a header file needs a definition/declaration that exists in another > header file, it should include that other header file. any other > approach is wrong. And if you try this in the real world for a project as large as the kernel you will die in a mass of unresolvable dependancy loops. The

Re: [PATCH] Include from .

2007-06-08 Thread Robert P. J. Day
On Fri, 8 Jun 2007, Alan Cox wrote: > > i don't see how fixing the C file solves this issue. the header > > file clearly refers to "struct ktermios", for > > which it won't have a definition unless it includes > > . > > Anyone using tty stuff will include to get all the > stuff needed an that

Re: [PATCH] Include from .

2007-06-08 Thread Robert P. J. Day
On Fri, 8 Jun 2007, Alan Cox wrote: > > > >Risk - which compile combinations does it actually show up with and > > > >is it not better to fix the C file in question not grow the > > > >horribly tangled include deps further > > > > > > struct ktermios; /* that's it */ > > > > um ... what does that

Re: [PATCH] Include from .

2007-06-08 Thread Alan Cox
> i don't see how fixing the C file solves this issue. the header file > clearly refers to "struct ktermios", for which it > won't have a definition unless it includes . Anyone using tty stuff will include to get all the stuff needed an that includes asm/termios.h which includes asm/termbits.h

Re: [PATCH] Include from .

2007-06-08 Thread Alan Cox
> > >Risk - which compile combinations does it actually show up with and > > >is it not better to fix the C file in question not grow the > > >horribly tangled include deps further > > > > struct ktermios; /* that's it */ > > um ... what does that mean? consider this snippet from > : Which

Re: [PATCH] Include from .

2007-06-08 Thread Robert P. J. Day
On Fri, 8 Jun 2007, Jan Engelhardt wrote: > > On Jun 8 2007 16:30, Alan Cox wrote: > > > >> > >> The header file should include > >> to pick up the definition of "struct ktermios"; otherwise, you run the > >> risk of seeing: > > > >Risk - which compile combinations does it actually show up with

Re: [PATCH] Include from .

2007-06-08 Thread Robert P. J. Day
On Fri, 8 Jun 2007, Alan Cox wrote: > On Fri, 8 Jun 2007 09:29:12 -0400 (EDT) > "Robert P. J. Day" <[EMAIL PROTECTED]> wrote: > > > The header file should include > > to pick up the definition of "struct ktermios"; > > otherwise, you run the risk of seeing: > > Risk - which compile combinations

Re: [PATCH] Include from .

2007-06-08 Thread Jan Engelhardt
On Jun 8 2007 16:30, Alan Cox wrote: > >> >> The header file should include >> to pick up the definition of "struct ktermios"; otherwise, you run the >> risk of seeing: > >Risk - which compile combinations does it actually show up with and is it >not better to fix the C file in question not

Re: [PATCH] Include from .

2007-06-08 Thread Alan Cox
On Fri, 8 Jun 2007 09:29:12 -0400 (EDT) "Robert P. J. Day" <[EMAIL PROTECTED]> wrote: > > The header file should include > to pick up the definition of "struct ktermios"; otherwise, you run the > risk of seeing: Risk - which compile combinations does it actually show up with and is it not

[PATCH] Include from .

2007-06-08 Thread Robert P. J. Day
The header file should include to pick up the definition of "struct ktermios"; otherwise, you run the risk of seeing: include/linux/tty_driver.h:142: warning: 'struct ktermios' \ declared inside parameter list include/linux/tty_driver.h:142: warning: its scope is only \ this