Re: Cancel-safe without a thread?


<BubbaGump> wrote in message
news:v4t0g2pl07b5f2uhohpqt751olvfjebj78@xxxxxxxxxx

I was thinking. When you put it like that, we may have common
enemies. It sounds like you want drivers to be perfect, and so do I.
In those terms, the type of people who I have problems with are the
ones who, when some potential problem is pointed out that they might
have missed, act like it's a hassle and want to ignore it because all
the tests they've run in their offices so far work fine. At that
point, it's not just incompetence. It's also deliberate negligence.
I don't know why people who don't want to ignore details don't choose
the type of work that's done in high-level languages where more
mistakes are caught in a safer and more obvious way.


The high-level language is a good analogy, I consider much of the driver
boilerplate to be like a high-level language:

1. The code is "generated for you" in the boilerplate it is by
Microsoft, in the language by the compiler.

2. People are used to the code, since in the boilerplate is seen over
and over again, and in the language it is a common compiler construct.

3. If there is a problem, lots of people are tackling a fix, if it is
the boilerplate developers talk about it on forums like this, and offer
fixes. If it is the compiler, people file bug reports and ask for fixes /
workarounds.


--
Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
http://www.windrvr.com
Remove StopSpam from the email to reply



.

Relevant Pages

  • Re: Extra reference?
    ... boilerplate does actually mean one and only one way ... each driver must implement this on its own. ... The status quo isn't necessarily right. ... glory, but less glory, less pain when the glory fades away. ...
    (microsoft.public.development.device.drivers)
  • Re: 2.6.11-rc2-mm1
    ... > To compile this driver as a module, ... Could we kill this boilerplate? ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • Re: [2.6.19 PATCH 7/7] ehea: Makefile & Kconfig
    ... >> +config EHEA ... > People usually add the following boilerplate: ... > To compile this driver as a module, ...
    (Linux-Kernel)
  • Re: New to USB, Is this possible?
    ... Of course you could argue that this constitutes writing a ... driver - just in a particular high-level language. ... a driver install file on it since I won't have access to all of the ...
    (comp.arch.embedded)
  • Re: [2.6.19 PATCH 7/7] ehea: Makefile & Kconfig
    ... To compile this driver as a module, ... +config EHEA ... People usually add the following boilerplate: ...
    (Linux-Kernel)