MyTI99.com Chess Game

[+9640News]

I need to ask a question regarding the chess game on MyTI99.com.  I am in the process of converting Stuart's Browser over to MDOS, and the only issue I am having with any of the conversion is the display of the chessboard and pieces.  All other graphics on Myti99.com work.

With the chess board, I get the links drawn on the right side of the display.  When it goes to draw the actual chessboard, the lower right two squares and the two chess pieces are drawn correctly.   The next chess piece to the immediate left of the first two, the top left of the 4 square grid is drawn with a partial chess piece, but then it stops displaying the rest of the chessboard.  Nothing more on the drawing.  Something has caused it to stop rendering the rest of the page/chessboard.  I have reviewed the page buffer where the page is stored and I see no immediate issue.  I think the code breaks for some unknown reason in the <8pclr: tag unless it is not getting for some reason the complete board.

 

If it is possible, can the mastermind behind the chess game for the TIPI provide me a page dump of the chess board with the pieces displayed?  That way, I can load it into the source code and chase down whether there is some underlying issue in retrieving the complete html page, or if I am dealing with another issue?

 

Thanks if you can help.  I don't know the User's Atari name to flag the user directly.

Beery

 

[Stuart]

The chess game is the masterpiece of Corey Anderson - @ElectricLab. Looks like he isn't visiting AtariAge too often now.

 

Have you tried these page to see if they work? Does everything seem to work apart from the chess game?

 

www.stuartconner.me.uk/tibrowser/z.htm
www.stuartconner.me.uk/tibrowser/z_ti_test_2.htm 
www.stuartconner.me.uk/tibrowser/z_ti_test_3.htm 

www.stuartconner.me.uk/tibrowser/valley.htm

www.stuartconner.me.uk/tibrowser/elite.htm

 

The chess board should be rendered from the top left to the bottom right. If only the bottom right squares are being rendered, I suspect there may be a problem receiving the data for the higher squares. Can you post what you're seeing in the page buffer?

 

You might be able to capture a page dump by opening a DOS prompt (assuming you're using Windows) and entering:

 

curl -A "TI09.4" "myti99.com/cg.pl?..."

 

... where the second bit in quotes is the URL you see on the top line of the browser. Not sure if it will work or not - not sure if the chess game is setting a cookie with your username and password in.

 

Might the TIPI Raspberry PI be able to capture outgoing HTTP requests and the data returned?

 

Edited February 10, 2023 by Stuart

[+9640News]

I hope this shows up right.  It looks like the whole page was captured as things are null terminated. 

 

All the pages above appear to render fine.  The elite webpage stays in a loop I am unable to break out of.  I need to add a keyscan somewhere in the code to test for F9, likely in the tag that tells it to load another page.

 

It is just the chess page, and the MAME dump is shown below.  The buffer for it starts at >6000 and extends up.  I use memory from >0400 to >3FFF, then from >6000 and up to around >ECFF is available for the program.  I have workspaces beginning at >ED00 which are subject to movement later to onboard ram.  I use keyboard XOP's and moved references to GPLWS and the >8374, >8375, and a few other address to labels instead of specific memory locations for video and sound.  I presently have commented out the specific joystick references as I do not have things on the Geneve configured for a joystick which would use the keyboard XOP's to get the joystick responses.

 

 

 

[+9640News]

4 hours ago, Stuart said:

 

Might the TIPI Raspberry PI be able to capture outgoing HTTP requests and the data returned?

 

Thinking about it, I should be able to do a block memory save from CPU adding a hotkey press.  Not sure how much I can gain from that, but it is worth a try.  As far as what the screen looks like:

 

[+9640News]

I was just able to do a memory dump from within the program, and save the page out.  Looking at the code disassembled code as all DATA statementes, I have several terminator strings before the completion of the first 4K with there being a significant amount of additional html code afterwards for the chess board.  I need to investigate the download page code to see if for some reason the recvmsg code could be passing an empty string back and appending a null byte which is providing a termination character, in the middle of the page. This might be an issue where the Geneve is running faster in MDOS than the GPL environment and why it is not seen.

 

Either way, the above screen display is always reproducible.

 

More testing tomorrow.

[+9640News]

I have resolved the chess issue.  After something you said, it made me realize I could do a memory dump instead of looking at MAME's DEBUG memory.  From there, I was able to search and find a rendering termination byte.  When i looked at it with fresh eyes this morning, I saw right after the chain termination byte, a URL link had been added.  That observation resulted in me looking at the end of the source file where I discovered I had improperly moved the LKBUFF that had created the issue.

 

So, as far as I know, everything works now.

 

Beery

[ElectricLab]

On 2/10/2023 at 6:21 PM, 9640News said:

I hope this shows up right.  It looks like the whole page was captured as things are null terminated. 

 

All the pages above appear to render fine.  The elite webpage stays in a loop I am unable to break out of.  I need to add a keyscan somewhere in the code to test for F9, likely in the tag that tells it to load another page.

 

It is just the chess page, and the MAME dump is shown below.  The buffer for it starts at >6000 and extends up.  I use memory from >0400 to >3FFF, then from >6000 and up to around >ECFF is available for the program.  I have workspaces beginning at >ED00 which are subject to movement later to onboard ram.  I use keyboard XOP's and moved references to GPLWS and the >8374, >8375, and a few other address to labels instead of specific memory locations for video and sound.  I presently have commented out the specific joystick references as I do not have things on the Geneve configured for a joystick which would use the keyboard XOP's to get the joystick responses.

 

 

 

Hmm, I see some whitespace in that 99ml that could be optimized out.