Re: [Simulavr-devel] unit test framework added

[Klaus Rudolph]

Hi Thomas,

I use version 1.6.0 and it is quite easy to "install" the framework
itself. Simply download and unpack. No compiling, no installation. Only
put the sources to the right ( configuarble) place. So I prefer to only
support the actual version and not some older ones from personal distros.

> Maybe you can send me your makefile as starter.

It is already part of the repo... ups... now it is :-) Makefile is part
of git ignore, so I can't see it. After autotools support was added it
should be moved to Makefile.in

> Do you plan to write some google tests more?

Yes!
After I stopped using simulavr for a long time there were a lot of
changes which brakes a lot of functionality and makes the tool unusable
for me. The 1.0.0 version is really fully broken. Wrong irq handling, no
support of multi core simulation, some hardware parts are broken and no
data values anymore in the trace output and lot of ICE faults during
normal run. It is a bit hard to see that the own code moves away from
the project and is exchanged with buggy replacement :-(

After the multi core support seems to work again and some other things
could be fixed in a few days, I hope we get the tool useable again.
Espacialy for multi core simulation I have to test the new scheduler. I
am in hope it works :-) I have no idea why the old code was replaced...
but I will not discuss the things, I only want a working tool :-)

Thats the reason I want to place a lot more tests there to avoid again
such a bad situation.


Actually i have a big problem:
I have 3 versions of simulavr.

1) "My" old one, which is working and correct but only supports a few
devices. I did a lot of changes there to fix bugs. But these fixes are
not part of the git repo.

2) A version which I have started based on the 1.0.0 version. I replaced
the scheduler, AvrDevice::Step and a lot of other code nearly complete
back to my old code to get that version work. That is not part of git
and will never be part of them.

3) I started regression of actual git head and run into lots of problems
again. As a result I commited some changes the last days, but my old
real world examples could not run under the actual head. There are some
things still broken but I could not solve all the problems until now.

I hope that I could get the git head to run the next days, but there is
a hard time limit. I realy could not spend to much time on it, because I
must get my project to work and actually it looks that I will do it with
my version 1). :-(

But for each bug I could fix I would also provide a unit test (if possible).

A lot of work :-)

So any help is really welcome!

Regards
 Klaus