asente@adobe.com (Paul Asente) (02/14/91)
In article <190*diane@drao.nrc.ca> diane@drao.nrc.ca (Diane Parchomchuk) writes: >Are X intrinsic routines and structures from X written to be robust enough >to adhere to the RISC architecture requirements? Yes. I'm not so sure about Motif however; 1.0 at least tends to spray "Fixing unaligned accesses" all over the place. -paul asente asente@adobe.com ...decwrl!adobe!asente
diane@drao.nrc.ca (Diane Parchomchuk) (02/14/91)
Because of my limited knowledge of C/RISC the ordering of parameters was explained to me locally as one of sizing, obviously from comments on the net this was not correct interpretation of the problem, alignment is *different*. Its just that (whine) us poor individuals who have learned C on UN-RISCY machines are not aware of this situation (end of whine). Are X intrinsic routines and structures from X written to be robust enough to adhere to the RISC architecture requirements? To be fair to the Decspurts they exactly explained the situation in terms of alignment -- I however don't know why compilers can't deal with this correctly so any alignment problem may cause inefficiency but at least work. Is this part of Ansii C? Diane
dbrooks@osf.org (David Brooks) (02/14/91)
asente@adobe.com (Paul Asente) writes: |> In article <190*diane@drao.nrc.ca> diane@drao.nrc.ca (Diane Parchomchuk) writes: |> >Are X intrinsic routines and structures from X written to be robust enough |> >to adhere to the RISC architecture requirements? |> |> Yes. I'm not so sure about Motif however; 1.0 at least tends to spray |> "Fixing unaligned accesses" all over the place. Ahem. Turned out to be a bug in the Ultrix regular expression routines (regcmp or regex; I forget which). Said bug was fixed in Ultrix 4.0 (if you don't have that, rebuild to use re_comp and re_exec instead). -- David Brooks dbrooks@osf.org Systems Engineering, OSF uunet!osf.org!dbrooks "Home is the bright cave under the hat." -- Lance Morrow