home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.perl
- Path: sparky!uunet!decwrl!deccrl!news.crl.dec.com!rdg.dec.com!decvax.dec.com!decvax!evans
- From: evans@zk3.dec.com (Marc Evans)
- Subject: Re: perl on Alpha AXP
- Message-ID: <1992Nov23.123008.16135@decvax.dec.com>
- Sender: usenet@decvax.dec.com (Usenet News System)
- Nntp-Posting-Host: akrisis.zk3.dec.com
- Reply-To: evans@zk3.dec.com
- Organization: Synergytics
- References: <1992Nov20.063505.4001@ulrik.uio.no> <2B0DC2D4.20446@news.service.uci.edu>
- Date: Mon, 23 Nov 1992 12:30:08 GMT
- Lines: 54
-
- In article <2B0DC2D4.20446@news.service.uci.edu>, iglesias@draco.acs.uci.edu (Mike Iglesias) writes:
- |> In article <1992Nov20.063505.4001@ulrik.uio.no> tleifsen@astro.uio.no (Torben Leifsen) writes:
- |> >
- |> >Has anyone ported perl to Digitals Alpha AXP/OSF-1 yet?
- |> >
- |>
- |> I used the hints that Marc Evans posted a couple of months ago to get
- |> perl 4.035 to compile and pass the tests. However, it doesn't work
- |> quite right. When I try to run Tom Christiansen's makewhatis script
- |> (part of his excellent man package), perl seg faults compiling the
- |> script (it even dies if I use perl -c makewhatis). I've included the
- |> output of dbx 'where' below. I'm not sure how to proceed from here,
- |> so if anyone has any suggestions please let me know.
-
- The problem with your core dumping is the OS, as I recall. I don't seem to
- have this problem on by current systems, which are likely running newer
- OS bits then you have ;-). I believe that there was some problems in the
- *alloc libc functions back in the BL9 timeframe...
-
- |>
- |> One other note: nm dies trying to output the symbols in libdbm.a
- |> during Configure. I took -ldbm out of the library list and nm worked
- |> fine and perl linked with no problems.
-
- There was actually a pretty nasty bug in nm'ing archive files (.a files)
- until recently (BL10.2). The result is core dumps when the conditions are
- *right*, and even worse is mis-information when those same conditions are
- slightly different. This too has been fixed.
-
- >From: john@sekrit.WPI.EDU (John Stoffel)
- >
- >As I remember, it was only a single line, and something along the
- >lines of changing a long to a time_t, or something like that. If you
- >can't figure it out, let me know and I'll see what I can do.
-
- The time_t problem will NOT surface if you use the -std1 option to cc. This
- is being caused by perl sources trying to define a function return type
- which is in conflict with that of the function prototype in the include files.
- By using the -std1 option the compiler runs in strict ANSI-C mode and
- therefore defines __STDC__ which the perl sources use so that the function
- is not declared.
-
- It is my *opinion* that you should try to make use of the -std1 option in all
- of your invocations (not just perl) as this will aid greatly in the portability
- of your code to POSIX.1, XPG/3, and other *standard* systems.
-
- - Marc
- --
- ===========================================================================
- Marc Evans - evans@zk3.dec.com | Synergytics (603)635-8876
- WB1GRH - WB1GRH@W2XO.#WPA.PA.USA.NOAM | 21 Hinds Ln, Pelham, NH 03076
- ---------------------------------------------------------------------------
- Unix and X11 Software Consultant
- ===========================================================================
-