This is the mail archive of the
dwarf2@corp.sgi.com
mailing list for the dwarf2 project.
Re: Rewrite of PROPOSAL 001213.1 re default location
- To: brender at gemgrp dot zko dot dec dot com (Ron 603-884-2088), Jason Merrill <jason at redhat dot com>
- Subject: Re: Rewrite of PROPOSAL 001213.1 re default location
- From: David B Anderson <davea at quasar dot engr dot sgi dot com>
- Date: Tue, 16 Jan 2001 07:57:02 -0800 (PST)
- Cc: BRENDER at gemgrp dot zko dot dec dot com, DWARF2 at corp dot sgi dot com, david dot weatherford at sun dot com
- References: <01011216161249@gemgrp.zko.dec.com>
- Reply-To: David B Anderson <davea at quasar dot engr dot sgi dot com>
Jason writes:
|> At this point x has two values, corresponding to the two assignments.
|> Both are valid and correct, and neither should be considered to hide or
|> take priority over the other.
|
|I disagree. We must associate a source position with /*break*/, and it
|seems intuitive for the debug info to choose whichever of x1 or x2 are
|currently associated with x at that source position. After all, the user
|is almost certainly interested in the value of x as mentioned in the code
^^^^^^^^^^^^^^^^
|displayed.
Apparently whether you take this position depends on whether you think
a debugger should
hide the extra stuff introduced by optimization,
making the program look like there was none and it
was simply the original unoptimized app
or
show all that is going on.
Both of these positions have been taken in the literature, so
I don't think we can say one is right and the other wrong.
|This example would produce a complicated location list, potentially
|alternating several times between the two registers, but it would be no
|more complicated than the line number information.
davea@sgi.com