Rendering of test page - some inconsistencies.
Cross-platform BBS (ANSI) Terminal
Brought to you by:
deuce
Menu
▾
▴
2 Attachments
#182 Rendering of test page - some inconsistencies.
Sorry, another one :)
Howdy, not sure if you are aware of the "Engineering Test Page" - I have it on my BBS as page *999#, and I think it is correct - but for testing, I've used telstar's test page *91# to be sure.
The code for the test page can be seen by the link in test.txt (its a long link), which will bring you to a videotext editor.
I think there are some issues with the reveal/conceal rendering, and I've added boxes to around it.
The small box (top), the "05" (in red) is missing, and the bars before 05 should be:
green blank yellow blank 05 (red) <-- reveal off
when reveal is on, the blanks become lines of cyan and white respectively.
At the bottom of the screen, all the text should be yellow.
Discussion
Anonymous
Thanks again!
Yes, I'm familiar with it, and use it as a normal test, but I don't have what "right" looks like seared into my brain yet. :D
Just pushed a commit to fix it, and kicked of a rebuild of the nightlies... as usual, let me know if this fixes the problems. Hopefully this is the last Prestel regression.
I also just changed the font for Prestel mode. Though you might be interested.
Howdy, nice work - its great, looks like it's all fixed.
A question, when I got a copy of the engineering test page, the bottom line has start box 0x0b and end box 0x0a codes. When I render that page from my BBS, it results in a "?" being rendered - but with telstar its a space.
I dont remember what those codes were used for, and I've been trying to figure it out - I find a lot of specs referencing start/end box, but no functional description of what it does. Have you come across it? How do you handle those codes?
Those allow the TV signal to pass through the terminal window, and are not supposed to do anything on a Prestel terminal... a ? is absolutely the wrong thing.
Last edit: Stephen James Hurd 2025-01-22
Yeah, I've seen references to them being used to overlay a TV image as well (and I guess not obstructing the TV picture - ala subtitles). So it probably does make sense on a terminal they are ignored. Its just odd they are referenced start/end box on some videotex specs, but not described.
That test page is also why I had to remove the Alpanumer Black and Mosaic Black controls frim SyncTERM... rendering that page properly requires ignoring them. :(
I did discover somewhere "versions" of the videotex spec, that covered when mosiac black (as well as double width) where used. In prestel (and viatel which was in Aus), the spec at the time didnt have them. I think those were used on teletext as the spec evolved and videotex died?
There's a decent photograph of the test page on real hardware in this thread...
Looking at my copy of the test page, there are no box characters in it.
Ditto for the version you linked when exported as Viewdata.
Grabbing a capture from your BBS, those are actually sent on the wire as ? characters... it looks like something is doing some kind of translation there. Which of the five ?s are which byte?
Hmm.. I must be doing that somewhere (assuming sync doesnt) - I'll need to look.
Normally with with zdnet, I export saved pages as "bin 1", and when I did that, a hexdump shows 0x0a and 0x0b where the ?s are rendered.
000003a0 6c 08 46 6c 61 73 68 03 2a 0b 0b 42 6f 78 09 53 |l.Flash.*..Box.S|
000003b0 74 65 61 64 79 18 47 6f 6e 65 0a 0a 3f 16 5e 7f |teady.Gone..?.^.|
I dont think I messed things up, but I'll double check.
Also, 0x0b is "Active Position Up" while 0x0a is "Active Position Down". Start box is \x1bJ and end box is \x1bK those I do have in my copy, and they correctly render as spaces.
Hmm.. I have definitely seen start/end box as 0x0a/0x0b - and it does overlap of active user keystrokes (where those are up/down as you described). I've never seen frames store movement chars (I think that idea was before its time).
Oh! I bet you have them as 8-bit data (where they would be 0x8a and 0x8b), but the high bit gets stripped since Prestel is a 7-bit protocol. Presumably the thing doing the stripping is noticing the high bit is set and swapping for a ?
I'll see if I can't construct an 8-bit engineering test page and see what happens... I'm not sure how I expect SyncTERM to behave to be honest. :D
Yup, I've just redownloaded the engineer page I sent you - if you show reveal codes on the website, you'll see that start/end box codes are used there, and when exported they export as 0x0a/0x0b.
Now to figure out why I'm sending them as ? :)
Sorry for the run around, I found it. I had those chars on my "what do I do with these list" and render a "?". I'll fix that, and I think you can consider prestel terminal emulation complete and awesome :)
Ah, yeah, I'm talking about the formats that go over the wire, so the Viewdata format from there (maybe the BBC RAM dump too, not sure on that one)
The Viewdata format is a 7-bit format that uses escapes for codes from 0x80 to 0x9f, but there's likely another one that uses the binary values (and SyncTERM also supports that).
Anyway, closing this issue... it's been an exciting day in Prestel emulation land, nailed down some very specific quirks on IRC with a guy that is willing to run stuff on his Model B for me, so we explored some dark murky corners of the standards.
So there's a lot of misuse of terms and confusion around Videotex/Teletext/Viewdata/etc.
They're pretty much all "Videotex", (But Viewdata was genericized away from having any meaning) and teletext is just "Videotex over television". By far the best and most complete spec is the ITU-T Recommendation T.101 (www.itu.int) which includes everything from Prestel to NAPLPS.
I believe Viatel was 8-bit, so closer to Minitel than Prestel, but I could be wrong, that's just off the top of my head.