Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!tut.cis.ohio-state.edu!uc!shamash!hare!ddh From: ddh@hare.cdc.com (dd horsfall x-4622) Newsgroups: comp.sys.mips Subject: A Loader question Keywords: ld Message-ID: <23898@shamash.cdc.com> Date: 28 Jul 90 00:10:10 GMT Sender: news@shamash.cdc.com Reply-To: ddh@dash.udev.cdc.COM (dd horsfall) Organization: Control Data Corporation, Arden Hills, MN Lines: 34 OK, friends and neighbors, your opinion is solicited as follows: I have a library, lib4.a, containing ~12 main programs and a small plethora of subprograms. According to the documentation (aha, there's the man's first problem!), I ought to be able to load main program N, named pgmN, and its requisite subprograms, with (1) ld -o pgmN -u pgmN lib4.a thus producing executable file pgmN. All _I_ can get, however is Usage: ld [options] files [...] (2) using 'cc' or 'f77' as the driver, inplace of 'ld', same thing (3) using pgmN_ as the Symbol Name, same again (4) using '-L. -l4' in place of explicitly naming lib4.a, same again (5-7) various combinations of above, just for completeness, same again. The language is f77, although the diagnostic certainly doesn't indicate a language dependency. Machine is R3xxx, OS is 4.-not-sure. [ Local advice to date has been, "Aw, Dan, you're _still_ trying to write stuff for a cyber", with *bonehead* clearly implied. ] One solution is obviously to keep 1 (smaller) library, sans main-pgms, and ~12 pgmN.o files, thus force-loading the proper main program and using the library for subprograms only. However, since I ran across '-u', that seemed to be worth looking at. I also hoped -u was the path whereby block data modules, which aren't explicitly called, might be included in the load. O Net, where have I gone astray? -- Horse [ Dan Horsfall +1-612-482-4622 + Control Data Corporation ] [ internet = ddh@dash.udev.cdc.com + 4201 Lexington Ave North ] [ uucp = TBD + Arden Hills MN 55126 USA ]