Windows upgrade questions

Steve G1XOW

Hi All,

As an MCSE, I would like to understand what the route cause is for the problems with CAT that happens when people are upgrading from W8 to W10, or indeed from W10 to later builds of W10 for that matter.

Are all the CAT virtual drivers (DRV and INF) files being deployed exactly per the Microsoft developer prescribed methodology? if so why is a refresh / upgrade not picking them up correctly?

Are FRS going to fix this problem, or are we to assume that we have to go through the rigmarole and significant disruption of uninstalling and reinstalling all SSDR versions with every new Windows release/build?

I dont know of a single comms/network/vsp that needs this kind of treatment, so I am sure everyone would like to know what is happening with the bug.

Also, keep in mind that M/S has said that there will be no new versions of Windows. This implies lots and lots of piecemeal upgrades every few weeks/months from now onwards.

We really need to know what to expect as we move forward to V2.0 regarding this unhappy relationship of SSDR and Windows 10 please

73 de Steve G1XOW

Steve G1XOW

edited content due to forum web app losing half of my text. Please re-read.

EA4GLI

My bet is driver signing issues..... but interested as well in an official answer. There are others that have not had issues with windows 10 so it might not be just a software issue.

From what I have gathered UPGRADES to windows 10 are way more unreliable than a FRESH INSTALL of Windows 10. It would be interesting to find out if the persons facing issues had done the upgrade or new install.

Peter K1PGV

Moving drivers in an upgrade scenario is complicated, and is never *guaranteed* to work. This is doubly true for anything that has a serial port number, due to the ancient, creaking, semi-broken nature of assigning serial port numbers. Add it the fact that there's an associated tool tray app... And good luck. Just for good measure, due to the small size of the installed base I'm sure that nobody at MSFT tests upgrading these drivers and provides specific work-arounds to any problems discovered the way they do for mass market peripherals. Yes... It "should" all "just work"... And in many if not most cases it does. But, unfortunately, any individual user's mileage may vary. Peter K1PGV P.S. Just to establish bona fides: I write, and teach people how to write, drivers for Windows... I'm a 15+ year Microsoft MVP in this space , a regular consultant to MSFT, and spent several years consulting to the Windows driver kits team (who develop the WDK).

Steve G1XOW

Peter,

Nice university-grade theory, but what I'm looking for is specific diagnostic evidence not just conjecture as to why something complicated may happen.

Obviously this kind of stuff could happen, but it is very far from normal, so we do need to get to the bottom of it.

If we can get enough people to like this topic then we may get some answers instead of just a **** and unacceptable work-around.

Duane_AC5AA

I just "upgraded" for the 2nd time from Win 7 on my shack computer yesterday, and once again have to go back and try to recover the Flex configurations. If I remember correctly, one clue might be due to "ghost ports" and I plan to search the community for Tim's post a while back on that problem.

Peter K1PGV

Hmmm...  Not a university-grade theory.  I'm telling you how it is in the real world.

Perhaps I wasn't clear: I design and write drivers for Windows every day of my life... and have been doing this for "a while" now. I was part of the team at Microsoft that designed how drivers are upgraded from one version of Windows to another. I personally know the guy who wrote the installer code for setup. I may not know much, but I know Windows drivers and operating system internals.

So, when I give you my professional opinion when it comes to Windows drivers I'm not provided a conjecture.  I'm telling you definitively how it is: There is never any guarantee that you can upgrade a driver in place from one version of the OS to the next.

Could the Flex team spend a lot of time and perhaps make the upgrade process a bit more predictable for some number of the vast possible combinations of upgrade scenarios?  Yes, I wouldn't be surprised if they could. Would that guarantee a problem-free upgrade?  I'm sorry, but it wouldn't.

Peter
K1PGV

Walt - KZ1F

@Steve, I am not articulate enough in current Window driver technology to have framed the question as well as you have. I do, and have had, the suspicion the issue was not Microsoft's. However, to thread the needle between what you said and what Peter said (welcome back, Peter), my take is if the technology is that buggy to implement properly, it's too buggy to use in production.

Be it DAX or CAT, simply provide sample applets to developers on how to natively connect to SSDR, and simplify, simplify, simplify. This is how standards become standards.. This is an opportunity for FRS to own SDR interfacing.

Mike va3mw

Flex does recommend pulling off SSDR completely before upgrading. While this might not answer your question, this is the work around. I had to do that last week. Tim has it well documented. Yes, it would be great if the drivers came online like they should, however, my guess is that it is a time to diagnose it issue. Maybe you MSCE types could spend some time looking at a failing system and see what state those drivers are in. Just a thought. Mike va3mw

Walt - KZ1F

Seems to me the genesis of the problem is trying to cram new technology into the format of old technology and it doesn't appear to be working well. I suggest embrace the new technology, own it!.

Bill W2PKY

Hello Peter, upgraded from Win 7 Pro to Win 10 Pro. One issue is after SSDR has been running for a few hours WSJT-X looses the transmit connection to DAX. I have to restart both DAX and the WSJT-X application to regain audio connection to DAX. Any ideas? Thanks;

Bob G W1GLV

I just upgraded to Win10 after much apprehension. The upgrade went smooth. The reinstall action of SSDR v1.8.3 went without a hitch. All systems are working perfectly no complaints. Here's what I did to make things go smoothly. I closed DAX and SDRCat. Removed FlexVSP and cleaned out the registry entries for FlexVSP, removed SSDR and removed all registry related entries. This was done using Advanced Uninstaller Pro V1.2. Restarted the computer. Installed Win 10 and SSDR software.

Steve G1XOW

@Pter,

Maybe I didn't make myself clear either. I have spent 32 years as a computer engineer, and over 10 years working for MS. I qualified as MCSE on Server 2000 !

My firm solves around 20-30 software issues every day, many of which are W10 related. So yes, I spend my time at the down and dirty end of systems which claim to work yet they somehow don't do what the promised.

Frankly it is disappointing that someone claiming to teach drivers to MS is prepared to propagate "it happens" as a reason to accept bad driver deployment and assume it has to be that way. Maybe 20 years ago that was acceptable when BSOD was the order of the day and a shrug of the shoulders was the best we could do.

These days this stuff should in fact "just work". If not then is it so bad to ask why not?   Along with the commercial benefits, the cost of propriety systems is that FRS do have to accept the onus of fault-free deployment.

Peter K1PGV

OK, Steve.  I bow to your obviously superior experience and knowledge of driver development and installation procedures.  Clearly you know best.

I'm anxious to hear what you discover as the root cause of these upgrade failures, and what you implement to avoid them. That will indeed be a benefit, not just to Flex, but to the entire Windows driver development community.

Good luck!

Peter
K1PGV

Peter K1PGV

@VA3MW:  Right.  Remove and re-install.  This is, admittedly, a PITA with the rolling-upgrades (what MSFT is calling "Windows as a service").  But this is the only way I've ever know for any driver update to be guaranteed to work properly... and often involves a small  manual configuration step as the last step of the process (like choosing the port number or selecting which sound card is the default). In fact, for some major Windows version upgrades, the Windows installer actually does a new install of the driver in the background. The problems come down to (a) the order in which things are installed, and (b) what state information the previously installed driver and its paired application(s) keep -- there's really no way for the Windows upgrade installer to know.

@KZ1F: Yes... You are correct. Absent these hideous legacy serial port issues, driver upgrades *generally* work pretty well. The way the serial port number database is managed is absolutely archaic... Which is fitting for an absolutely archaic technology.  Again, this assumes that the installed driver and associated app(s) don't save specific state in unusual locations on disk or in the Registry. SOME of that can be moved... most can't because what's important or useful can't be identified.

The reason that most commodity peripheral driver upgrades work without incident is because either (a) they don't store state, (b) they don't come with proprietary paired applications, (c) there is special accommodation made for their upgrade within the installer by MSFT (you'd be surprised how many of these cases there are), (d) they don't use serial ports, (e) some combination of a through d.

Peter
K1PGV

Bill W2PKY

Thanks for the AU Pro tip
 

Steve G1XOW

@Peter,I have no argument with you personally, only that your approach is somewhat defeatist long before we have even had the dialogue about the root cause of the problem, or how it might be addressed. As I.T. professionals we should always push hard against the “it happens” response to problems which only undermines user confidence and tends to suggest that all driver problems are in the O/S, which as we know they are not!

I am sure You would take a different view if you also faced perpetual reinstallations of SSDR every few months in tandem with O/S build releases.

My business support 4,700 PCs, and about 280 Servers. We deal with lots of legacy hardware, plenty of XP era kit, along with many many RS232 and VSP drivers for interfacing to ticketing machines, barrier control systems, epos systems, conveyor and sortation systems, barcode scanners etc. etc.

In almost 12 months of W10 rollout I have only seen two (2) driver issues. One was a very old USB to Ethernet adaptor from a company I have never heard of, the other a WIFI adaptor where no suitable driver existed. They both went in the bin - job done for £20. 
So there is good evidence to feel that there is not a generic driver instability in W10.  An observation which my friends at Siemens Industrial systems also agree with.

Note to all: this is likely to be an ongoing problem, not just when upgrading from say W7 to W10 but for all future builds of W10. 

I want to ask FRS directly to see what they think the problem is. And, yes I am also talking to the Industrial Systems support team of Microsoft Germany too. I hope to pull some conclusions from both sides, even if that means we just have to accept “it happens”. At least then we will know where we stand and make buying/upgrade decisions from an informed standpoint.

...trying desperately not to own a dog yet still finding the need to bark myself!

73 de Steve G1XOW

Michael Coslo

I suppose you wil get many answers, but what about those of us that have not had a problem? I put my radio though it's paces, using SmartSDR on a Dell Optiplex 960, an Imac running Windows 7 via bootcamp, and DogPark SDR on the Mac side of the mac.

The only problem I've ever had was a missing font on my Dell that took a few minutes to fix.

To my way of thinking, I am a prime candidate for problems. 


And this isn't to brag - I am genuinely interested in why some do, and some don't.

I do know I've worked with a lot of people who have had audio issues in W10, often they need to update their sound car drivers. The trick is getting them to believe you. They've run the troubleshooter and it says it's okay when it actually wasn't. 

As an MCSE, do you have any ideas regarding the disparity?

Michael Coslo

Oddly enough, along with not having the audio problems, after a few times of doing the uninstall, I just install the new version, remove the old version shortcuts, and run with that. I'm beginning to think I'm living in a dreamworld, maybe lucky?


So why do some have the problem, and others such as myself not? As noted in another post, I have one "normal W10 setup, a W7 running in Bootcamp on a Mac, and a Mac version of software to control the Flex. Therein might lie a big clue

G8ZPX

Michael,

I too am genuinely interested the the problem and why it should be happening to some SSDR users. There are problems with both CAT and DAX drivers.

My concern is not for me, reinstalling in my game is very normal and nothing to be stressed about. My concern is for those people who are not so Windows savvy and have to expect problems on a perpetual basis until it gets addressed.

So I am raising the flag for others that do not just have to accept it. So far the subject has been swept under the carpet, but we all deserve an answer from FRS on this.

I do have several hunches on this, but was waiting for some feedback from FRS before drawing conclussions. Observations so far suggest a problem of re-assimilation of non-core data sets for the VSP driver after a Windows refresh. In other words machine state data is being stored outside of the normal prescribed (but limited) driver structure and therefore does not get recognized by Windows upon a refresh.

I have never seen this with problem with the 3rd party VSP solution for my legacy Flex 5000a.