Device Upgrade Problem With Shifted Keys and RMIR

WagonMaster · Post by **WagonMaster** » Sat Sep 12, 2009 8:57 pm

I've run into a problem with RMIR (v1.96) that may be a documented shortcoming that I just don't know about.

Basically, I was using a device upgrade that uses some shifted keys. But every time I uploaded (with RMIR, under Linux) to the remote using this device upgrade, all of the unshifted keys worked fine but none of the shifted keys did!

I scratched my head for a long time on this, trying all sorts of things, and finally just tried to do the upload with 'IR.exe'. Voila! All of the shifted keys were now working perfectly.

I never noticed it at first, but in retrospect, I now see that the keymoves needed to implement the shifted keys were not getting loaded in RMIR when the device upgrade that implemented those shifted keys was being loaded!

Is this a known design shortcoming (or a known bug) or am I doing something wrong? If it's a bug, I'm having a hard time believing that I'm the first person to run into this, so I suspect I'm missing something here.

Please enlighten me, oh wise ones of this list.

Regards,
Bill

gfb107 · Post by **gfb107** » Sun Sep 13, 2009 8:47 am

RMIR is not functionally complete. Basically it is complete enough for my setup, as to date I am pretty much the only one using it. I don't use shifted keys, so that's why this bug is still there.

Thanks for reporting it, I'll look into fixing it. Happy to have another RMIR user, especially one who seems interested in contributing to the coding effort.

WagonMaster · Post by **WagonMaster** » Sun Sep 13, 2009 1:46 pm

gfb107 wrote:... as to date I am pretty much the only one using it.

Not to contradict the guy who's offering to investigate the problem

, but I'd be surprised if that's really the case.

Despite being intimately familiar with the serial library now, I don't have a lot of experience with the JP1.x user tools. I try to use RMIR first because I like the idea (natively cross-platform) and the integration. But sometimes (as expected, given that it is, as you say, "not functionally complete") I see weird things and, if I'm thinking straight, I'll just switch to 'IR.exe' and try that, reporting what I find if I think it's a bug and not my own ignorance.

gfb107 wrote:Thanks for reporting it, I'll look into fixing it.

Thanks, Greg... I appreciate whatever attention you can afford to give it!

gfb107 wrote:Happy to have another RMIR user, especially one who seems interested in contributing to the coding effort.

I'd try to assist more with the coding (at least fixing bugs), but I'm not really a Java guy and every time I try to figure something out in RM/RMIR, I just cannot seem to do it. In part, I get stymied by that mass of files! When I get past that, I'll load something that looks like the right source file, browse around, feel lost, repeat that 'N' times, and eventually give up.

Bill

gfb107 · Post by **gfb107** » Wed Sep 16, 2009 12:31 pm

I finally carved out some time to look into this. I've been unable to reproduce bug.

Please provide more information about what you are doing. Which remote? What keys? What device upgrade? If you have an IR or RMIR file I can use as a starting point please upload it to the Diagnosis Area and provide a link.

WagonMaster · Post by **WagonMaster** » Wed Sep 16, 2009 5:08 pm

gfb107 wrote:I finally carved out some time to look into this. I've been unable to reproduce bug.

OK... Thanks for taking the time, Greg.

gfb107 wrote:Please provide more information about what you are doing.

I will definitely do that either later tonight or maybe tomorrow. The only reason I didn't go into more detail on my initial post was that I somewhat expected this to be a "not yet implemented" response, especially since it occurred with 2 totally different remote controls. Because of that, I didn't spend any time writing up an extensive test scenario posting, but I have good notes and I'll do that soon.

Thanks again!

Regards,
Bill

WagonMaster · Post by **WagonMaster** » Wed Sep 16, 2009 7:21 pm

gfb107 wrote:I've been unable to reproduce bug.

Please provide more information about what you are doing.

OK, I took the time to document this in more detail for this post....

This was all done with RM/RMIR v1.96 on Linux (Slackware 12.1). I'm using my updated Linux 'jp12serial.so' (v0.14) library, but that shouldn't be causing the problem that I'm seeing.

I'm being very explicit in the description of the steps. I know you don't need such detail, Greg, but I wanted to be precise in case anyone else wants to try this out.

Steps taken to duplicate the problem:

Load the 'Philips_3576H_160GB.rmdu' device upgrade (from the forum) into RM.
Select the "Functions" tab and then the "Clean up" pushbutton to get around the other problem I posted about (corruption of the "Function.__.hex=__" lines in the saved RMDU file).
Change the target remote from RS 15-1994 to RS 15-135. Ignore/close the "Lost Function Assignments" dialog.
Save this device upgrade into a new RMDU file. The RMDU file looks fine.
Shut down RM and run RMIR.
Download from RS 15-135 remote or load up an RMIR or IR flash data file with a reasonable setup. Delete any existing device upgrades, macros, and "learns" as needed. I simply loaded the "virgin" 15-135 flash memory setup (downloaded when the remote was new).
On RMIR's "Devices" tab, select "New". On the "Device Upgrade Editor" window that pops up, select "Load". Load the device upgrade that you saved to disk a few steps back. Click "OK".
Since this device upgrade uses code "DVD/1010", go back to the "General" tab and enter that "Setup Code" for the "DVD" "Device Button" entry.

I don't think I noticed this before, but the "Raw Data" tab shows a bunch of bytes that are colored red. I don't know what this means, but I suppose it's telling me something important.

Notice that the "Key Moves" tab is empty, which is not what I'd expect given that "Output" tab on RM shows the 10 keymoves needed for the shifted keys.
Save this setup as a new RMIR file then upload it to the RS 15-135.

Aside: I frequently get an error dialog after upload saying "Upload verify failed: data read back doesn't match data written.". I have been unable to identify a pattern, but it often seems like shutting down RMIR, opening it back up, re-loading the RMIR file, then doing the upload makes the upload succeed.

I've been meaning to look into this myself sometime because I figured it would be a good introduction to RMIR code. But I haven't had a moment to look into it so far.
When I test the freshly uploaded RS 15-135 on the DVR, all of the unshifted keys work fine but none of the shifted keys work.

When I first saw this problem (6 days ago), I had done all of this with a URC-8820 remote as well and it failed similarly.

I didn't repeat this part tonight, but if one copies the device upgrade bytes (which include the shifted keys implemented as key moves) from the "Output" tab of RM, pastes that into 'IR.exe' in the usual manner, and then uploads that to the remote, all of the keys (unshifted and shifted) work as expected.

As near as I can tell, somehow the shifted keys aren't making it into the "Key Moves" tab of RMIR when the device upgrade is loaded into RMIR. My understanding of all this may be flawed, but that's what it seems like to me.

To be thorough, I re-enabled the original (v0.13) serial library, re-ran RMIR v1.96, re-loaded the "virgin" RS 15-135 load, and re-applied the saved device upgrade from the saved file. I still don't see anything in the "Key Moves" tab of RMIR.

I've uploaded a file named 'shifted-keys-problem.zip' which has the modified RMDU and the RMIR I used to upload to my RS 15-135.

I hope this description and zip file help you duplicate things, but if I've inadvertently omitted anything, please let me know.

Any guidance is appreciated. Thanks!

Regards,
Bill

mdavej · Post by **mdavej** » Wed Sep 16, 2009 8:32 pm

Bill,

I think you've just stumbled on a very poorly formatted file. Things go awry in your step 4. The output really isn't fine. When I converted your file as-is in RM to the 15-135, none of the shifted keys but one survived the conversion. The button descriptions made it for those shifted keys, but not the actual functions. So it appeared fine, but the keymoves weren't really in the output.

I think the source of the problem is the function names. Many contain a large number of spaces followed by an asterisk. I think this is throwing off the parsing somehow. I've never seen anyone else name functions in such an odd way.

Here's what I did to clean it up. After that, all the shifts work fine.
- Open the original RM file formatted for the 1994
- Go to the functions tab and click Clean Up
- Edit each function description that has an asterisk, removing the asterisk and all the spaces in between
- Now pick the 15-135 and the shifted functions are preserved

I don't know if Greg needs to make any changes to handle such an unusual case.

WagonMaster · Post by **WagonMaster** » Wed Sep 16, 2009 11:10 pm

Thanks for your inputs, 'mdavej'. I had composed a lengthy reply describing my confusion with your post but by the time I went to post it, you changed your post!

Fortunately, your changes clarified some of my confusion about what you were saying. But unfortunately, I'm still very confused. Allow me to explain....

mdavej wrote:Things go awry in your step 4. The output really isn't fine. When I converted your file as-is in RM to the 15-135, none of the shifted keys but one survived the conversion. The button descriptions made it for those shifted keys, but not the actual functions.

I'm not seeing this at all. I just checked the file I uploaded and all 10 of the shifted keys have proper definitions for the "Function.##." lines they reference. What are you seeing, specifically, in the RMDU file that I uploaded that looks damaged or corrupted or missing? I just don't understand and expect I'm missing something here. As long as you performed my step #2 (which clearly you must have since your sequence describes it too), everything (at that point) is still OK, no? Please clarify. Thanks!

Regards,
Bill

mdavej · Post by **mdavej** » Wed Sep 16, 2009 11:55 pm

The corruption I was talking about occurs if you change remotes in RM without doing the clean up. I now see that I glossed right over your statement that you DID do a clean up.

I admit I didn't get as far as trying your zip file. Now that I did, I can't get the shifted functions to load into RMIR either, even though they're in the RDMU.

So we're back to square one.

However, if Greg wants to streamline the troubleshooting a little, he could simply load the RDMU from your zip file and immediately see that the keymoves don't come in.

In case you haven't looked at it, there are also clues in the rmaster.err file, but I don't know what to make of them.

WagonMaster · Post by **WagonMaster** » Thu Sep 17, 2009 12:14 am

mdavej wrote:I now see that I glossed right over your statement that you DID do a clean up.

OK, good -- that completely clears up my confusion.

mdavej wrote:In case you haven't looked at it, there are also clues in the rmaster.err file, but I don't know what to make of them.

Thanks for the reminder. I hadn't actually looked at it, probably because I was thinking of this problem more as an oversight which RMIR wouldn't have any debug output about as opposed to an actual detected/reported error condition. I'll try it again and take a look in the 'rmaster.err' file for more clues.

By the way, I think those asterisks in the original author's RMDU file indicate functions that don't have a button on the Philips DVR's original remote control. I have no idea why he/she included all those superfluous spaces, though.

Bill

gfb107 · Post by **gfb107** » Thu Sep 17, 2009 5:39 am

Don't know if you guys have read the RMIR: Prototype IR function in RM thread, but RMIR treats key moves a bit differently that IR. Specifically:

key moves that
have the same device type and setup code as their bound device button and

the device type and setup code match an installed device upgrade
will automatically become part of the device upgrade, and not be shown as a keymove

RMIR considers these keymoves to be part of the device upgrade, and as such can only be seen and manipulated from within that device upgrade.

Of course the key moves should still be installed into the remote when uploaded, it's just a question of how they are treated in the UI.

I'll try out your test scenario and let you know what I find.

WagonMaster · Post by **WagonMaster** » Thu Sep 17, 2009 8:30 am

gfb107 wrote:RMIR considers these keymoves to be part of the device upgrade, and as such can only be seen and manipulated from within that device upgrade.

A few days after I had my problems with RMIR (2 documented in threads and a few others, like the seemingly incorrect "upload mismatch" warning), I actually did read all 12 pages of that thread, somewhat casually over the course of 2 days, IIRC. Obviously, though, the importance of that line slipped by me somehow and didn't register as germane to my problem.

Thanks for the enlightenment.

gfb107 wrote:I'll try out your test scenario and let you know what I find.

Thanks, Greg... I appreciate it!

By the way, as reminded by 'mdavej', I went back and checked the 'rmaster.err' log file after duplicating the problem and saw nothing obvious to clue me in to the source of the problem.

Also, somewhere along the way in my JP1 "studies"

, I seem to have missed the explanation of the meaning of red-colored bytes in RMIR's "Raw Data" tab. Can someone please fill me in on that?

Regards,
Bill

gfb107 · Post by **gfb107** » Thu Sep 17, 2009 8:35 am

The red bytes in the Raw Data tab are bytes that have been changed since loading a file/downloading from the remote.

WagonMaster · Post by **WagonMaster** » Thu Sep 17, 2009 9:07 am

gfb107 wrote:The red bytes in the Raw Data tab are bytes that have been changed since loading a file/downloading from the remote.

Ah, thanks! Seems obvious in retrospect. I was worried that it meant something slightly more ominous. I'm learning....

Bill

wnewell · Post by **wnewell** » Thu Sep 17, 2009 9:51 am

I'm seeing the same as you in an upgrade I'm working on but I didn't even try an upload since I noticed the keymoves didn't make it into rmir.

And for Greg, There's probably a lot more people that would prefer to use rmir over ir.exe, but I'm at least one more.