Update: Response to Serotek’s mirror driver document

« The Joys of Windows 2000 in the Year 2007

Mar 19

Update: Response to Serotek’s mirror driver document

After reading the original document from Serotek on mirror drivers, I created my response and posted it on our blog March 14th. However, once Serotek read my reply they drastically updated their document based on the information I posted but they did not keep the original document. If people come in to this cold and only read their updated response (because the original has disappeared) it would be very confusing. The readers would also miss how one sided, biased and inaccurate this document was. I hate to put such a document on our GW Micro web page but I feel I must to give people the correct background of where this document started. If you are interested in reading the original document you can click on Serotek’s original document.

I’m not going to take the time to respond to their updated document in details as we have our heads down for CSUN this Wednesday. But I am happy to see they are now owning up to using API hooking. I wish they would have addressed the concerns about API hooking that I outlined. None of that was addressed. The biggest being that Microsoft doesn’t want API hooking to be used.

I also read a comment on my original blog saying we are all hyping this up and using scare tactics etc. and should be focusing on UIA. This may seem obvious but once you get into the technical details I would ask that person why? If Window-Eyes 6.0 supported UIA today why would this be better? The answer is it wouldn’t but it is cool to say we should be supporting the latest and greatest and very easy to say we aren’t because it is too hard or we are to entrenched or we are too busy defending the approach we have. Nothing could be further.

I would like the person that commented to demonstrate how UIA (managed or unmanaged) would help applications that are being used today? The answer is it wouldn’t. In fact, Microsoft’s original response with Vista is there will be no API hooking and there will be no video driver support meaning UIA/MSAA is it. However this would leave the thousands and thousands of legacy applications out in the cold. Absolutely no support. Only through our lobbying did Microsoft then do the necessary work to give a good legacy solution and that solution is mirror drivers. So UIA is not the end all be all.

Also, I would like the person that commented on this to explain what applications, major or minor are currently using UIA today? Vista currently does not ship with any UIA applications nor are there any major shipping applications for the main stream.

As I mentioned in my original blog, we’ll use what we need for the times. I have no doubt we’ll be using UIA in the near future. The commenter also didn’t mention that Microsoft will be very shortly removing the restriction of being managed on the client side. They have already removed the managed restriction on the server side but didn’t have time to remove it on the client side (the AT side) before Vista shipped. Knowing that there are no applications that ship with UIA and that the managed restriction would be removed shortly I wonder again why this person thought UIA was so much more important for today?

Again, I agree tomorrow but without fully knowing the state of Windows at this point in time it seems a bit easy to sit back and make wild claims and assumptions. Rest assured, Window-Eyes will use the best approach for the times to get the necessary support to today’s applications.

4 comments

§ Doug ® said on : 2007-03-20 @ 11:01
After reading my own blog I realized I might not have been as clear as I could have been regarding UIA. Microsoft has obviously implemented UIA within the operating system and standard controls. However, my point was at this moment in time, there is no advantage of using UIA to support these controls over our existing implementation. Obviously this will change as Windows Presentation Foundation (WPF), formerly code named Avalon is used more, and applications are written to directly support the advantages of UIA but for now there is no immediate need for direct UIA support. Rest assured it will be coming however. I hope this helps clear up my point.
§ Josh de Lioncourt said on : 2007-03-20 @ 18:27
Having read the original document previously, I agree there updated version was drastically modified. However, much of it is still just as mis-leading as ever. The two most striking things to me, and they would've been even if I hadn't already been aware of most of this, is their continued reference to DCM or Mirror driver technologies as antiquated, and yet they call API hooking both an early approach and the way of the future. That aside, many of the other arguments they use against mirror drivers could quite easily be applied to their API hooking approach, especially where concerns system stability and security. And, I haven't heard anyone call mirror drivers a permanent sollution. As Windows evolves, so will the products that give access to it. They are quite as naive to suppose their approach is so prepared for future versions of the OS, as they accuse the other developers of being.
It seems to me this edit of the original document was clearly an attempt to save face, and was lacking tremendously in its effectiveness therein.c
§ Mike Duigou said on : 2007-09-19 @ 18:55
Why not use an RDP loopback instead of DCM or a mirror driver? Supporting RDP is going to be critical anyway as remote and virtualised sessions are increasingly common.

(For this entry a link to UIA would have helped)
§ Tomas Finnøy said on : 2007-11-06 @ 07:21
"If Window-Eyes 6.0 supported UIA today why would this be better? The answer is it wouldn’t but it is cool to say we should be supporting the latest and greatest and very easy to say we aren’t because it is too hard or we are to entrenched or we are too busy defending the approach we have. Nothing could be further."

And what application developer would spend time supporting UIA when there are no screen-readers/accessibility tool developers willing to put in the effort themselves? I believe that the accessibility community should work together and demand support for these techs by implementing the client part asap. It would look rather silly to demand such support from application developers, when we are not willing to do it ourselves.

I guess I was a bit too quick when saying that both of you were defending your choice of ancient technology. You DO state that it is only(?) in your interest to use mirror drivers to support legacy applications.

Backward compability IS important, but so is the utilization of new and better technologies. I feel that UIA won't be supported by the community while discussions like these are taking place.

BOTH, as you say, are for legacy applications, and BOTH, in my opinion, should be considered in a phase-out stage.

The same argument is valid for managed vs unmanaged.

Comment feed for this post

GW Micro Blog