[Ohrrpgce] SVN: pkmnfrk/2290 That's strange, it compiled for me...

Ralph Versteegen teeemcee at gmail.com
Sun Sep 28 03:43:16 PDT 2008


2008/9/26 James Paige <Bob at hamsterrepublic.com>:
> On Thu, Sep 25, 2008 at 03:14:06PM -0400, Mike Caron wrote:
>> James Paige wrote:
>> >On Thu, Sep 25, 2008 at 10:56:18AM -0700, subversion at HamsterRepublic.com
>> >wrote:
>> >>pkmnfrk
>> >>2008-09-25 10:56:18 -0700 (Thu, 25 Sep 2008)
>> >>38
>> >>That's strange, it compiled for me...
>> >>---
>> >>U   wip/subs2.bas
>> >
>> >The windows version compiles without -g -exx by default, so although it
>> >compiled for you, you had potential memory corruption at runtime.
>> >
>> >It broke the compile for me because on linux I compile with -g -exx by
>> >default.
>> >
>> >(actually, I guess the -exx part is the only part that matters for
>> >bounds checking, the -g is unelated)
>>
>> Oh. I see. It's actually been many moons since I dug deep into the
>> batch-file hell I forged, and forgot that -exx (and -g?!) weren't included.
>>
>> Which really makes me wonder if there isn't some reason to add them now...
>
> Here are the reasons that I can remember (and my memory is fuzzy)
>
> 1) Bigger executable filesize
Much bigger actually
> 2) Fear of slight slowdown (I think TMC even profiled it)
Massively slower actually. About 3 times slower, I think. Due to null
pointer checks in the blitting code.
> 3) Run-time error messages are impossible to display on Windows.
> Application just suddenly closes. Contrast Linux which prints filename
> and line-number where the error occured toi the console
Under Windows you have to compile with -s console to get the error
messages printed, though of course you still need to run from the
console. And for some reason the messages are also sometimes missing
line number information.
I was going to create a debug build with batch files and gdb scripts
to do everything for bug reporters automatically. The scripts are
already in wip/misc.
> 4) I never previously noticed that -exx enabled compile-time bounds
> checking in addition to the run-time error checking
That is really silly. What kind of BASIC doesn't care about error checking?
>
> ---
> James



More information about the Ohrrpgce mailing list