This screen mode is not suitable for displaying the desktop
Chris Hall (132) 3559 posts |
Entry and EXIT are macros used for … etc. Many thanks for that. Almost there. It now runs with no errors (see below) but just gives the result Width=0 height=0. Still puzzled but will look more carefully tonight.
|
Jeffrey Lee (213) 6048 posts |
You’ve got some Entry/EXIT issues with HWP_Send – the Entry is missing and it’s using EXITs that corresponds to looksee. |
Chris Hall (132) 3559 posts |
You’ve got some Entry/EXIT issues with HWP_Send Many, many thanks, so I had. It now responds with 1920 and 1080. Gosh! ScrHelp utility updated to version 0.03. Still a few warts as BASIC sometimes crashes (I suspect I may be pushing my luck with stack space). Also only works on Pi (need some error traps to allow it ignore Pi-specific stuff for other platforms). |
Steve Drain (222) 1620 posts |
I have realised that was not a good example, because
|
Steffen Huber (91) 1953 posts |
Is there a problem with 1920×1200 on any RPI model? BeagleBoard, certainly. PandaBoard, sometimes. But the RPi? |
Steffen Huber (91) 1953 posts |
If you configure hdmi_group=2, you can use hdmi_mode=77 which should get you 2560×1600. Some have even succeeded with 4K video output, although that seems to limit refresh rate to 15-20Hz, which is not supported by most displays. |
Chris Hall (132) 3559 posts |
Program still does not work at reading the EDID on this one. Assume you are using version 0.03? Could you please run the following code and send me the result? I can then update the programme to work correctly (I only have a few monitors to test it on).
|
Chris Hall (132) 3559 posts |
Ran that, can not send the result, as have no where to send it to. send to fromdave (at) svrsig (dot) org – many thanks STR$ now corrected in line 70 above. |
Chris Hall (132) 3559 posts |
The EDID data you sent contains the following information: Due to a bug, only the first of the first four is reported, i.e. 640×480 @ 60Hz – look out for version 0.05! Due to the EDID data not following the spec the last two listed resolutions are not included where they should be (in bytes 38 to 53 of the block starting at buf%+28) where it specifies the x-resolution, y-resolution and frequency as up to 8 2-byte data sets. They are included as detailed timing information which takes a bit more decoding and as the spec just says it duplicates the easier to decode information I ignore it. Only the last three resolutions decoded are shown in the window. |
Chris Evans (457) 1614 posts |
Chris, for clarity can you confirm that your results above, were using data from DavidS’s native 1920×1200 monitor? We did some test work on about 20 monitors some years ago for EDID on the a9home, about half the EDID files were wrong in one way or another. When EDID first came to RO5 we were told monitors are now nearly all correct… |
Chris Hall (132) 3559 posts |
Chris, for clarity can you confirm that your results above, were using data from DavidS’s native 1920×1200 monitor? It is from his data. The ‘standard timing information – up to 8 2-byte fields’ is not used (set to 0101) in his data despite the spec requiring it and there are two detailed timing descriptors (1280 wide and 1920 wide) where the ‘vertical active lines’ are shown as &2D0 and &21C (i.e. 720 and 540) where the latter should clearly be &438 (i.e. 1080) or &4B0 (i.e. 1200). Version 0.06 now reports what the EDID data say (including, at risk of duplication, info from the [first four] detailed timing descriptors, which are listed in decreasing preference order) – I can’t correct errors in the data! |
Chris Hall (132) 3559 posts |
Isn’t raspian correct for a native 1920×1200 monitor? |
Jeffrey Lee (213) 6048 posts |
540 * 2 = 1080. You’re almost certainly failing to decode the DTDs for interlaced modes correctly. David: Can you try posting the mode file created by *CreateModeFile? ScreenModes’ EDID handling isn’t perfect, but it will be a lot more complete than Chris’s code. |
Chris Hall (132) 3559 posts |
it will be a lot more complete than Chris’s code. Certainly will. I have read the EDID spec but only implemented part of it (as my utility is for information only). The EDID data are definitely wrong though in places for this particular monitor. I am decoding the DTDs as provided, the first of which (the preferred resolution) is given in bytes 54-71:
This gives 1280 × 540 interlaced 60Hz Edit: I mean 1280 × 720 non-interlaced 60Hz – I transcribed the flags as 9E vice 1E. I am not trying to decode the alternative SVD blocks for the available DTV formats as these data blocks are optional whereas the ones listed above are obligatory. |
Jeffrey Lee (213) 6048 posts |
&2D0 = 720. That looks like it should be a DTD for a standard 1280×720 @ 60Hz mode, except that the interlace flag is set, which would give it a (probably nonsense) resolution of 1280×1440. |
Chris Hall (132) 3559 posts |
&2D0 = 720. Well spotted. The second (and final) DTD is:
Which gives 1920 × 540 interlaced (a bit more credible as 1920×1080 is a standard mode). except that the interlace flag is set, which would give it a (probably nonsense) resolution of 1280×1440. Oops – post corrected above, I read the wrong line when transcribing. almost certainly failing to decode the DTDs for interlaced modes correctly Version 0.08 should have sorted this out. Many thanks. |
Clive Semmens (2335) 3276 posts |
I understand that the RPi (with recent RISCOS builds) can support 3840 × 2160 at 30 Hz – if anyone who’s actually doing that feels extremely generous I’d love to have an SD card image to give it a whirl. I’m afraid I find the idea of doing a RISCOS build myself too intimidating :-( |
Chris Evans (457) 1614 posts |
You don’t need a special build of RISC OS. Just enter the correct information in config.txt We’ve had 2560×1440 running. There was an issue with the pointer disappearing at 2048 but that was fixed about a year ago. |
Clive Semmens (2335) 3276 posts |
I’m still running RISCOS 5.21 (08-Jul-13) – last time I looked, anything more recent than that was only available as a nightly build, not an image. Or have I missed something, or failed to understand something? Or was the fix something that could be applied without a rebuild of the OS? |
Clive Semmens (2335) 3276 posts |
Apparently – and I can see where to get the nightly build ROM image.
Sadly, this doesn’t provide any information, nor a link to such information, about how one might go about doing this. I’m probably moderately adventurous, but I don’t have a clue where to begin. |
Steffen Huber (91) 1953 posts |
Open the !Boot dir on the SD card you boot from. Does it contain a file called “Loader”? If so, this is the FAT partition the RPi boots from. Double-click on the Loader file (which is presented as a DOSFS image file), find the riscos/rom file. Replace it with a a new RISC OS ROM image, and you’re done. Well, in theory. You might also need to update your !Boot sequence, and maybe the Raspberry Pi firmware files, because not all RISC OS versions are compatible with all RPi firmware versions. You can download the RPi firmware from here: https://github.com/raspberrypi/firmware/tree/master/boot If there is no “Loader”, you have to put your (micro)SD card into a cardreader to access the FAT partition to replace the files. After writing all that, I agree that we should have an easy-to-follow tutorial somewhere. |
Clive Semmens (2335) 3276 posts |
Many thanks. I’ll do a new SD card with the latest RC14 image first and hold onto my existing one (which I find does have Loader, however). |
Rick Murray (539) 13851 posts |
A few additional notes:
|
Clive Semmens (2335) 3276 posts |
Hmm. Perhaps I’m not adventurous enough. I think I’ll hang on for a new RC. It would be nice to drive my big monitor, but I can live with the 1600×1200 for the moment! I don’t have a lot of older, crashy software – the only things I use are Draw, NetSurf, CreateSEC, ChangeFSI, FontEd, Paint, Zap, BBC BASIC, and stuff I wrote myself, mostly in BASIC with just a few chunks of assembler for heavy data manipulation – nothing that’s likely to upset anything. Does RC14 (8-Jul-13) run in High Vector mode? It’s giving me no trouble – apart from not driving my big monitor! |
Clive Semmens (2335) 3276 posts |
Thanks, David. I suspected that, but wasn’t sure. Like I wrote, I think I’ll wait for a new RC before trying to drive the big monitor – what Steffen said looked manageable, but Rick has scared me off! It sounds like the kind of thing I’d take on if I had to, but not cheerfully! |