Re: [Ecls-list] long-long

2010-07-04 Thread Matthew Mondor
On Sun, 4 Jul 2010 15:30:35 +0200 Juan Jose Garcia-Ripoll wrote: > ECL supports "long long" whenever the compiler allows it. That's nice. > Regarding the other question, ECL just follows the C naming convention. > Expecting long, int, long long or any other integer type to have a certain > size

Re: [Ecls-list] long-long

2010-07-04 Thread Juan Jose Garcia-Ripoll
On Sun, Jul 4, 2010 at 8:48 AM, Matthew Mondor wrote: > I wasn't yet hit by this case as I had no 64-bit API using long long to > deal with (NetBSD lseek(2) using a 64-bit off_t (a quad_t, itself a > typedef to int64_t)). But long long as been in use for a long time in > the C world and it ideall

Re: [Ecls-list] long-long

2010-07-04 Thread Juan Jose Garcia-Ripoll
On Sun, Jul 4, 2010 at 5:19 AM, Pascal J. Bourguignon wrote: > Juan Jose Garcia-Ripoll writes: > > > Seth, can you try compiling this tiny piece of code? (...) > > On 32-bit debian lenny (686) (I'd need long-long support on there too): > (see below for x86_64 gentoo). (...) Seems everything

Re: [Ecls-list] long-long

2010-07-03 Thread Matthew Mondor
On Sun, 04 Jul 2010 05:19:39 +0200 p...@informatimago.com (Pascal J. Bourguignon) wrote: > Juan Jose Garcia-Ripoll writes: > > > Seth, can you try compiling this tiny piece of code? > > > > (FFI:CLINES " > > long long lseek64(int a, long long b, int c) > > { > > return 0; > > } > > ") I wasn't

Re: [Ecls-list] long-long

2010-07-03 Thread Pascal J. Bourguignon
Juan Jose Garcia-Ripoll writes: > Seth, can you try compiling this tiny piece of code? > > (FFI:CLINES " > long long lseek64(int a, long long b, int c) > { > return 0; > } > ") > > (FFI:C-INLINE (1 2 1) > (:INT :LONG-LONG :INT) > :LONG-LONG > "lseek64(#0,#1,

Re: [Ecls-list] long-long

2010-07-01 Thread Juan Jose Garcia-Ripoll
On Thu, Jul 1, 2010 at 11:50 PM, Seth Burleigh wrote: > Anyways, now i have to figure out how to get > ecl to recognize that i actually do have dffi installed... > > You mean libffi? Looking at config.log and why it fails may help. Juanjo -- Instituto de FĂ­sica Fundamental, CSIC c/ Serrano, 1

Re: [Ecls-list] long-long

2010-07-01 Thread Seth Burleigh
for some reason, it now compiles fine, it may have had something to do with the multiple ecl versions i have installed (i thought over each other, but apparently not). Anyways, now i have to figure out how to get ecl to recognize that i actually do have dffi installed...

Re: [Ecls-list] long-long

2010-07-01 Thread Juan Jose Garcia-Ripoll
Seth, can you try compiling this tiny piece of code? (FFI:CLINES " long long lseek64(int a, long long b, int c) { return 0; } ") (FFI:C-INLINE (1 2 1) (:INT :LONG-LONG :INT) :LONG-LONG "lseek64(#0,#1,#2)" :ONE-LINER T :

Re: [Ecls-list] long-long

2010-07-01 Thread Juan Jose Garcia-Ripoll
Ok, sorry, the previous email was wrong: I was looking at a too old version of CFFI. In the recent versions the code looks fine #+long-long (:long-long :long-long "long long") #+long-long (:unsigned-long-long :unsigned-long-long "unsigned long long") but there is a lin

Re: [Ecls-list] long-long

2010-07-01 Thread Juan Jose Garcia-Ripoll
Sorry I did not answer this question before. I inspected your configuration logs and ecl_long_long_t is defined. Thus :LONG-LONG should be in *features*. If :LONG-LONG is in *features* then the compiler must support this type and ECL indeed does. However, CFFI might not. Indeed if you look at CFFI'