Complete.Org: Mailing Lists: Archives: freeciv-dev: June 2004:
[Freeciv-Dev] Re: (PR#9020) optimization of map_get_tile
Home

[Freeciv-Dev] Re: (PR#9020) optimization of map_get_tile

[Top] [All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
To: jdorje@xxxxxxxxxxxxxxxxxxxxx
Subject: [Freeciv-Dev] Re: (PR#9020) optimization of map_get_tile
From: "Gregory Berkolaiko" <Gregory.Berkolaiko@xxxxxxxxxxxxx>
Date: Sun, 20 Jun 2004 04:28:03 -0700
Reply-to: rt@xxxxxxxxxxx

<URL: http://rt.freeciv.org/Ticket/Display.html?id=9020 >

On Fri, 18 Jun 2004, Per Inge Mathisen wrote:

> > I disagree.I think it would be better just to enable a higher
> > optimization level in the compiler. Static functions can easily be
> > automatically inlined by an intelligent compiler, and we can hope it
> > will do a better job than we would at guessing which ones are the best
> > to inline.
>
> Well, that depends on whether we will actually use a higher optimization
> level or not, does it not?
>
> I have my doubts that everyone will agree that running all debug levels in
> -O3 is a good idea, and please do not make it so that --enable-debug=yes
> is unplayable.
>
> Maybe selectively picking gcc optimizations is a better idea than using
> -O3, though.
>
> >From the gcc manual: "-O3 turns on all optimizations specified by -O2 and
> also turns on the -finline-functions, -fweb and -frename-registers
> options. " About -frename-registers and -fweb it says they can "make
> debugging impossible", so it might not be a good idea to use them.

Even O2 makes debugging if not impossible but very hard.  Anyone who tried
to step through a function in gdb knows that the execution jumps between
lines erratically and the variables that you see do not necessarily exist.
So when I plan to do debugging, I compile with O0.

Also, for debugging having map_get_tile a macro would be quite bad, AFAIK
debugger doesn't recognize macros.

G.




[Prev in Thread] Current Thread [Next in Thread]