Thank you, Jan. I agree with your comments; at least to the extent that it
is often advantateous to invest in tools as you say. In the non-programmer
world, there are many tools that must be purchased in order to offer
products and services. Doctors, hair stylists, photographers and
construction workers must have tools that they often purchase themselves.
Programmers are less likely to think in terms of purchasing tools other than
a compiler and a computer with an operating system (usually not in that
order).
Except I don’t have $100. Eating at McDonalds is a luxury for me. I am
afraid that my cat will get sick or injured and I don’t have the money to
take care of her. Fortunately I don’t have a family that is dependent on me.
If I am able to develop something in a week that allows me to do what I need
to do then that will be very worthwhile, especially if I learn from it. I
originally learned about Windows and Windows programming (I was an IBM
“mainframe” type programmer long before the original PC was a glimmer in the
eye of IBM) as a result of developing a Windows printer device driver that
sent images to a HP LaserJet expansion board over SCSI using only the SDK
and DDK documentation to figure out how to do it; the internet was not yet
available for assistance. I will omit the details of that, but I can do that
stuff if I have documentation that gives me the necessary information.
I tried using IrpTracker because someone in the:
microsoft.public.development.device.drivers
newsgroup suggested I use it. What I really need is something that traces
the flow of data between the device driver and an application. I can do that
using a spy tool except I don’t know of a spy tool that shows the data and
especially not the data from asycnronous io. My intent was to pursue the API
spy solution further if I did not get something useful from the Microsoft
device drivers newsgroup.
I don’t know much about writing debuggers, but I might be able to solve my
problem by using a custom debugger. I think a debugger could be very useful
for another project I have in mind. If anyone wants to give me hints about
how to develop an API spy such as what I describe above, then it would be
appropriate for me to share the result. I already have some sample code in
my:
http://SimpleSamples.info
“Jan Bottorff” wrote in message news:xxxxx@ntdev…
>I use a commercial tool called BusHound sometimes. It’s I believe $600, but
> there is no way I could write a filter driver from scratch that does the
> same stuff as BusHound for $600. There is another too around which is
> similar, although the name escapes me at the moment. There is also the
> commercial tool IrpTrace, which I believe is $100, WAY cheaper than me
> writing a filter driver.
>
> Perhaps it’s just my way of thinking as a busy consultant, but if I can
> buy
> a tool for $500 to do what would cost my client $2000 for me to write, I
> get
> out the credit card and pay the $500. Realistically, I can’t write, test
> and
> document all that much functionality for $2000. Same goes for hardware,
> the
> expensive part of writing drivers (or any software) is DEVELOPER TIME.
>
> Actually, if you look at Microsoft’s financials, it says they generate
> $500k
> of sales per engineer, which implies the opportunity cost for a Microsoft
> engineer to be wasting time is around $200/hour. Many companies don’t
> translate their engineering time into sales quite as well as Microsoft;
> although using the cost of an engineer as $1000/day is an ok round number
> when looking at make vs. buy decisions. MANY companies don’t seem to
> understand this simple reality (little companies often understand it
> better)
> and will save a few dollars by not purchasing some development resource as
> they watch their engineering payroll costs skyrocket (or project progress
> move glacially slow). Software productivity studies say a developer can
> write around 10k lines of code a year (I tend to use 10k lines/year as
> sustainable long term, 20k lines/year as approaching long term burnout,
> and
> 30k+ lines/year as very serious burnout territory only possible for short
> periods), so at $1000/day, spending $2000 to write a utility gets you
> maybe
> 100 lines of code. Even though as a developer who can write more than 50
> lines code/day, over my 25+ year career as a software developer, these
> numbers fit reality surprisingly well (for high quality debugged code).
>
> - Jan
>
>>
>> I don’t see how to see the data from the requests captured by IrpTracker.
>
>
>