Test rigs, dev machines and psychic debugging



I got a cool new dev machine from my Boss as a sort of late Xmas
present a few months ago.

After all the fun of installing Delphi on it, and getting all of my
custom components to work to the point where I could rebuild a project
successfully, I ran into a bit of a lazy block where I couldn't be
bothered installing some of the additional software that it provided by
other parties, and that our software relies on.

In the past, my dev machine has been capable of doing anything that a
client machine would be able to do, but this new machine is much
cleaner, can't do a bunch of stuff, and I am pretty happy about it.

In the worst-case scenario, I can fire up a VM that is fully specced
and try to figure out problems that way. This happens rarely.

More often, I'll write a driver that emulates what the 3rd party stuff
is supposed to do, and leave it up to the testing dept to do end-to-end
checks.

I decided a while ago that my dev machine should not also be a test-
rig, so that leaves me with only psychic debugging as a way of tracking
down most bugs.

The thing that I have found most beneficial about this approach is
that it is way more productive/effective than anything I did earlier.

I've found that debugging most issues ends up being an exercise in
solving problems by thinking about them, in much the same way that you
might try to solve a Sudoku puzzle without looking at it.

Maybe there's also an element of being able to "hold a program in your
head".

Gone are the days of swearing at the IDE debugger and the way it
throws focus around indiscriminatley (using BDS2006 here). Instead, I
can just go to bed and sleep on the problem, secure in the knowledge
that in finding the bug, I might also find a good re-factoring
opportunity or algorithm improvement.

The systems that I work on are large and complex. There's nothing
really psychic about trying to understand every part of what the system
is doing simultaneously, but I do find that the restriction of not being
able to single step through what it is up to has helped me enormously.
It can, however, be a bit mind-bending.

The testers and customers think that I am psychic.

Do you work this way? Would/could you work this way?

Scout
.



Relevant Pages

  • Re: Visual studio 2005 and compact framework 1.0
    ... Thierry ... > Try installing the .NET Framework v1.1 SDK on your dev machine. ... > .NET Compact Framework MVP ...
    (microsoft.public.dotnet.framework.compactframework)
  • Re: Visual studio 2005 and compact framework 1.0
    ... >> Try installing the .NET Framework v1.1 SDK on your dev machine. ... >> .NET Compact Framework MVP ... >>> Thierry ...
    (microsoft.public.dotnet.framework.compactframework)
  • Go, software activation, go!
    ... Installing BDS2006 on my new dev machine. ... Calling the Borland India hotline using the 0800 number suggested for german callers, i get told, very friendly no less, that the are very sorry but cannot help me with product activation because they have are having problems with their database - could i call another day? ...
    (borland.public.delphi.non-technical)
  • Re: Test rigs, dev machines and psychic debugging
    ... After all the fun of installing Delphi on it, and getting all of my custom components to work to the point where I could rebuild a project successfully, I ran into a bit of a lazy block where I couldn't be bothered installing some of the additional software that it provided by other parties, and that our software relies on. ... In the past, my dev machine has been capable of doing anything that a client machine would be able to do, but this new machine is much cleaner, can't do a bunch of stuff, and I am pretty happy about it. ... I've found that debugging most issues ends up being an exercise in solving problems by thinking about them, in much the same way that you might try to solve a Sudoku puzzle without looking at it. ...
    (borland.public.delphi.non-technical)
  • Re: Speeding up emulator
    ... provides the computers at work I'll have to suffer with slowness. ... debugging on the device to see what happens. ... One other question about this "ethernet connection". ... >> a better dev machine next year. ...
    (microsoft.public.pocketpc.developer)