09/04/17 ROM breaks Screen config/Netsurf

Open a Netsurf window. Open Configuration – Screen. Move Netsurf window and enjoy the fireworks.

Haven’t checked any other app windows yet.

Edit: Seems to be just Netsurf. Tried Fireworkz, OPro, Pipedream, StrongEd, Zap & LaBella all of which seem OK.

Frederick

What version of Netsurf is that with and also what versions of support modules that it requires have you installed.

Also I assume you have updated your disc image as well?

Disc image is today’s. Netsurf is 3.7 (Dev C1 #4059). Support modules; AcornURI 0.26, Iconv 0.12, SharedUnixLibrary 1.14, Tinct 0.14

What’s happening is that after initial breakup of Netsurf’s window content any window passing over the Netsurf one leaves a trail but only after opening the screen config.

On a Titanium after running Configuration→Screen the NetSurf window goes ’Fireworks. Changing the mode to the current mode gets things back to normal or pressing F12, Return the desktop screen mode changes to 640×480 in 2 colours.
On a Raspberry Pi 3 a higher resolution is selected. Goes from 2048×1152 to 2560×1440. Colours stay the same. Netsurf remains unaffected.

Screen resolution here is 1024×768, the max & native for this Dell 1505FP. Beagleboard -xM.

Disc image is today’s. Netsurf is 3.7 (Dev C1 #4059). Support modules; AcornURI 0.26, Iconv 0.12, SharedUnixLibrary 1.14, Tinct 0.14

Try without the use of Tinct¹ by selecting OS based dither in the netsurf image handling config.

¹ May be irrelevant but I do recall strange things happening with Tinct use on releases a few years ago.

The Wimp probably needs fixing so that it will only refresh currentmode on Service_ModeFileChange if the desktop hasn’t been entered yet. currentmode is directly exposed by Wimp_ReadSysInfo 1 (it’s supposed to be the current Wimp mode), so once you’re in the desktop it should only really change on when an actual (Wimp) mode change occurs.

Try without the use of Tinct1 by selecting OS based dither in the netsurf image handling config.

Already selected.

The Wimp probably needs fixing so that it will only refresh currentmode on Service_ModeFileChange if the desktop hasn’t been entered yet.

Wouldn’t it be simpler to only refresh it when the task count changes from 0 to 1? I assume (though can’t actually see the line) it’s a call to Wimp_Initialise that triggers the initial mode change, in response to the *Desktop command starting a Wimp task.

Then there’d be no need to bother reading it when the module initialises, nor on Service_Reset, nor on Service_ModeFileChange.

But since I can’t see where the initial mode changes by just staring at it, I’ve no doubt missed something subtle and would need to stick a breakpoint in to see what’s going on.

But since I can’t see where the initial mode changes by just staring at it, I’ve no doubt missed something subtle and would need to stick a breakpoint in to see what’s going on.

Use the force, Luke! grep, Sprow!

The main place to worry about is Wimp_SetMode / *WimpMode. If the wimp isn’t active yet then it won’t actually trigger a mode change; all it will do is update currentmode, thereby selecting the screen mode that should be used when the Wimp does get entered. The boot sequence relies on this behaviour – the Configure.Monitor file is located in PreDesk.

Just to confirm, today’s ROM (11/12 Apr) has resolved the problem on this BB -xM.

Just to confirm, today’s ROM (11/12 Apr) has resolved the problem on this BB -xM.

I would like to congratulate whoever fixed this, and ask that the revisions page actually says who is responsible for updates – not to criticise failures, but to congratulate fixes for bugs and so forth.

Too often it says just ROOL, and the individual is anonymised.

I want to be able to say “Thank you, whoever!”

Too often it says just ROOL, and the individual is anonymised. I want to be able to say “Thank you, whoever!”

I agree with your comment.
If you dig deep enough into CVS I’ve often found a view of the sources that shows who wrote what, line by line with the authors initials but I often don’t seem to be able to find it.
I’ve just tracked down another view of the relevant sources here. That lists Rob Sprowson as the author of the above fix, though it did require a bit of digging included changing the view to sort by date.