Re: Linker: how to detect an unwanted redefinition of a libc routine (AIX 5.3)

From: Timothy J. Bogart (tbogart_at_frii.net)
Date: 11/14/05

  • Next message: Dohhhh: "Re: GXT255P in 7043-260 machine" 
    Date: Mon, 14 Nov 2005 12:55:45 -0700

    Gary R. Hook wrote:
    > Timothy J. Bogart wrote:
    >> Umm, do you have the patience/interest to contrast this with some
    >> non-unix that somehow doesn't have problems with namespace collisions?
    >
    > Well, I probably don't have the background with non-Unix OSes.
    > I can only comment on how Unix or Aix linkers behave, and couched
    > my response in those terms. I would readily agree that the issue
    > is rather global.

    Ah. Reasonable. The wording left me thinking you might have something
    to contrast.
    >
    >> Also, going back to the OP, you need not 'know' all the names in libc
    >> (or whatever), there is such a thing as letting the machine grep and
    >> tell you if the name you propose is used already ....
    >
    > This seems a wee bit unsophisticated.

    Please consider who posted this. 8-)

    Also, when compared to the concept of memorizing all the symbols, it
    might be regarded as downright elegant. 9-)

    > Why shouldn't the linker be
    > able to report on overrides of symbols that were found in archives?
    > Seems a perfectly reasonable request.

    Well, why not have the programming tools check for memory leaks, maybe
    automatically manage the memory for you? Maybe check for out of range
    input values ... gee, it seems the list goes on for what seems a
    reasomable thing to add. <grin>

    But more seriously, I wonder if the -E and -f options of the GNU linker
    don't address this. I am just to ignorant (or is that unsofisticated)
    to really be able to tell from the info blurbs - or come up with a test.
    >
    > FWIW I think that the first time anyone (everyone?) is bit by this
    > issue, they learn real fast how they can best avoid the problem
    > in the future. That's how I learned :-)

  • Next message: Dohhhh: "Re: GXT255P in 7043-260 machine"