[ros-dev] Re: [ros-diffs] [ion] 14047: Dispatching & Queue Rewrite II:

[Steven Edwards]

--- Alex Ionescu <ionucu at videotron.ca> wrote:
> This goes beyond debug information. This is reproduceable behaviour that 
> probably any driver developper out there knows. Checked builds are 
> builds recommended for testing your driver for bugs. If you call that 
> function with a Queue Object, you WILL see that assert line-by-line on 
> your screen. From that point on, one should stich his eyes out for 
> having seen it, and shoot himself for knowing this behaviour?

So fix a ReactOS driver to match this behavior and work under Windows and
then you will have justification for making related changes in the kernel.
 
> Notwithstanding that they cannot sue the project, and that they would 
> not sue you. This was a public comment to a friend... why would they sue 
> Steven when Alex said what he said? And yes, I cannot wait to be sued... 
> I can see the headlines -- Driver Developer sued for being aware of 
> Windows Assertion --. I hope they also go after Mark Russinovich for 
> having used the checked build to generate a tree of the Windows Source 
> code!!

The last time I looked Mr Russinovich was not try to make a replacement
for Windows but rather provide more information people wanting to use
Windows.

> If you aren't, then why am I always the one being targeted with such 
> comments. There are functions in ROS which are almost copies of their 
> binary versions. There are structures in ROS which look like clones of 
> the Windows ones (undocumented ones). There is functionality that was 
> directly reversed engineered so that it would be compatible.

Yes we have reverse engineered quite a bit but the question is what methods
are being used to reverse certain behavior. We cannot help but be compatible
with the structures in Windows and take any means needed to be compatible.
 
> Yet, nobody says a word; everyone goes after Alex for having a 
> conversation with a friend and mentionning a reproducible fact  in every 
> driver developer's life -- you do not KeWaitXxx on a Queue. 

> Probably as much as jumping on a guy who has written some of the highest 
> quality and most useful code in the OS for the fact he used public 
> information during an argument.

You stated the other day there were regressions that were only found by 
developing test cases. You would have a lot more good will from developers
on this project if you committed test cases for some of the patches you develop
and commit them to rosapps/tests or write a dummy driver to show the behavior
rather than than pointing to checked builds.

Thanks
Steven



		
__________________________________ 
Do you Yahoo!? 
Yahoo! Mail - Helps protect you from nasty viruses. 
http://promotions.yahoo.com/new_mail