[ts-gen] Compiling on an OS/X

[R P Herrold]

On Mon, 31 Dec 2007, William Lloyd wrote:

> XCode doesn;t seem to upgrade nicely.  Had to chop away directories
> to get it installed.
>
> There is no Ruby code called that I can see by the makefile, or at
> least up the point it's dying.

Makefiles are tricky, no doubt about it; as the top of the 
file notes, we have a local copy of Mecklenburg, "GNU Make", 
O'Reilly 3rd Ed. here to try to help us through the more 
ornate parts.  there was an intriguing 'make' behaviour, in 
that failed build intermediate files were not being deleted as 
we expected them to last month, and Bill spent a couple of 
days running that down.


The scripts [as with all scripts] (ending in .rb) are in the 
./bin/ helper directory, invoked by this line:

%.h : %.i               ;       @(bin/$(*F).rb $< > $@) || \
 	$(call you-need-ruby, $(@F))

which is making the .i prototypes (which yield the .h objects) 
as the compilation of the .c code proceeds, walking 'vpath' 
as follows:

vpath   %.i src

Those files are:

[herrold at centos-5 bin]$ ls *.rb
generation.rb  rdtsc_time.rb

This is implicitly called in by some .h dependency when it is 
solving inet.c in the example you posted.

I suspect that it you check the timestamp of cats.h and 
generation.h, they will be later than their peers, and that we 
have a unexpected case to code for on OSX/i686

The test in question is this one in rdtsc_time.rb:

 	uname -smp

and in looking at the logic, I see that we do NOT account for 
non 'powerpc' Darwin  in:

 	 def triple()

Would you be so kind as to advise of the value of:

 	uname -smp

and I'll write a patch for it.

Thanks for the report, as this will get us to additional arch 
coverage.

-- Russ Herrold

ps:  I am advised in IRC that it is as follows, but wish to 
confirm as to the host you (William) are building on.

10:19  orc_orc> jasonc: ping
10:19  orc_orc> or others -- I need on an Intel Apple, the string returned by:
                 uname -smp
10:20  jasonc> orc_orc: Darwin i386 i386
10:21  orc_orc> thank you -- building a patch atm for a ruby file
10:21  jasonc> wow that sounds... not fun ;)