1. Rob, convince me to part with my $88
That's registration & source code, not C translator.
Obviously I don't expect you to implement this for me (otherwise it
would be your $39 awaits..) but without proof-of-concept variants of
the following, either from your good self or the fine upstanding
members of the euphoria mailing list, I'll pass.
I suspect this is all going to fall on deaf ears because, although you
(Rob) are right about bind&profile being something unique and
sellable, trace is years out of date and frankly inadequate, but yet
not open source. That part I most urgently want to improve.
Wants:
1) Value trap: run the program until variable x in routine z of
include file y hits a given value; when also k=7 and j=5 etc.
(ie looking at ex.err, "what set that wongly?")
*Definately* require that to be x[327][4] set to {} when k[4]=7 etc.
(kinda rules out using types, which is never going to be an expedient
way of tracking down a simple bug anyway)
Note: 327,4,8,4,3 *are* literals/constants, very,very,very rarely it
might be handy to use eg x[last-first+1]; not worth it.
2) Debugger/trace: show me the value of eg itab[374]. Let me add
trap(condition) statements in my source code and let the debugger
resume over them. Obviously, before resuming, let me examine all the
local variables to check all is well. Allowing variable to be modified
before resuming is sometimes helpful.
Or, let me define such value traps/(condition)s when in debug mode (eg
by running a development version of exw.exe rather than the souped up
super-fast version normally shipped).
3) Replace the dos look& feel trace window with something more up to
date. For known reasons it is just too damn slow on w9x boxes. It
would be cool to have a tabbed view of variables, and set the
(development) exw.exe to show selected variables first/uninteresting
variables last. Allow user-defined tabs for "interesting" variables,
eg itab[374] (such settings saved between runs).
4) Some commitment on your part that if I modify 2.3 pd & ship you the
source diff then weekend/whenever; unsupported, you'll ship back
modified 2.3+profile. (Last major stable release only, so it is just
{compile,zip,post}). (So I am not force to have two different
development environments on my pc to get the full benefit of what I
paid for)
Pete
2. Re: Rob, convince me to part with my $88
Pete Lomax writes:
> 1) Value trap: run the program until variable x in routine z of
> include file y hits a given value; when also k=7 and j=5 etc.
You can do this easily already by defining a unique type for x,
and other ones for k, j. See the example in the reference manual.
Just turn on trace(1) when the condition occurs.
> 2) Debugger/trace: show me the value of eg itab[374].
I expect to do at least that much.
> 3) Replace the dos look& feel trace window with
> something more up to date.
We'll see. I can't commit to that.
> 4) Some commitment on your part that if I modify 2.3 pd & ship you the
> source diff then weekend/whenever; unsupported, you'll ship back
> modified 2.3+profile.
Sorry. You'd have to pay extra for that service.
Regards,
Rob Craig
Rapid Deployment Software
http://www.RapidEuphoria.com
3. Re: Rob, convince me to part with my $88
On Wed, 27 Feb 2002 20:43:56 -0500, Robert Craig
<rds at RapidEuphoria.com> wrote:
>
>Pete Lomax writes:
>> 2) Debugger/trace: show me the value of eg itab[374].
>
>I expect to do at least that much.
When? Can I help at all?
>
>> 3) Replace the dos look& feel trace window with
>> something more up to date.
>
>We'll see. I can't commit to that.
Ditto
Anything, factor out your prized, pride & joy, code, add a 3 month
time bomb to it, release only the binaries, & explain how to call it.
Whatever, something. Most of us here want to help, not profit at your
expense.
Pete
PS guess how much I just spent! I knew I had to.
