Path: utzoo!censor!geac!torsqnt!lethe!yunexus!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!hp4nl!delgeo!ooms From: ooms@delgeo.nl (Frank Ooms) Newsgroups: comp.unix.programmer Subject: Re: Question on improving link speed: ld -r vs. ar Message-ID: <2019@delgeo.delgeo.nl> Date: 28 Feb 91 07:55:40 GMT References: Organization: Delft Geophysical b.v. Lines: 29 In article montanaro@crdgw1.ge.com (Skip Montanaro) writes: > [ ... ] > 1. Given the restricted number of visible symbols in each module, is > it likely to improve link speeds? We have used this scheme for some time with a set of libraries and saw an improvement of only 10%. The total size of the libraries is in the order of 10 Mbytes, with about 6000 routines. The major drawback of this scheme is that is difficult to replace one routine with a private version for testing purposes, since that will result in multiply defined symbols. > > 2. How widely available is the -r flag to ld? I looked around a > number of machines on which we currently support our software, > and it seems to be on all of them (Sun, Stellar, HP 9000/300, > SGI), but I'm hesitant about future ports. > Well we have it (ConvexOS V8.1) on our BSD derived system. They claim POSIX compliability, so if this option is POSIX it will probably stay for a while. Cheers, -- /* Frank Ooms, +31 15-621554 * * ooms@delgeo.nl * * ..!mcsun!hp4nl!delgeo!ooms */