Re: [RFA,RFC] TARGET_ADJUST_BREAKPOINT_ADDRESS - patch 3 of 4

["Eli Zaretskii" <eliz@elta.co.il>]

> Date: Fri, 3 Oct 2003 23:43:53 -0700
> From: Kevin Buettner <kevinb@redhat.com>
> 
> I would also like to hear from Eli.  If it's not clear from my words
> above, I'll state again that I feel very strongly that a warning is
> more appropriate than a mere informational message which might be
> missed by the user.  The adjustment of a breakpoint *should* cause the
> user to be alarmed since the expected breakpoint behavior could be
> significantly altered.
> 
> So...
> 
> Michael S: Okay?
> 
> Eli: Comments?

First, let me thank you so much for taking the time to explain this
issue in so many words, let alone investigate the problems and write
the code and the docs.

As for the issue itself: I'm not 100% sold on your arguments, but
given my total lack of experience with any platform like the FR-V, I
don't see how I can in good faith avoid a total surrender! ;-)

So it's okay with me to have this as a warning, but let me make 2
suggestions that will allow me to swallow this frog and smile, too:

  . Please add some of the technical reasons for breakpoint
    adjustment into gdbint.texinfo (yes, I know that you wrote a
    patch for that, but the message I'm replying to has so much more
    info that it'd be a pity not to retain some of it for posterity).

  . Please add to gdb.texinfo a description of the warning message,
    whereby the exact message is cited followed by an explanation of
    why is this of interest to the user (e.g., that code rearrangement
    could have messed up line numbers, and the user should consider
    putting a breakpoint on other/additional lines).  Please add a
    @cindex entry to that description citing the beginning of the
    message text, like this:

     @cindex breakpoint address adjusted

    This way, a user who wants to understand the message needs to
    type "i breakpoint address adjusted RET" and read the node
    presented to her.

Thanks again for all your efforts.