Level-2 I/O does not work if file is opened with Level-3

[Kchula-Rrit]

In my ramblings with Level-2 and Level-3 I/O I found that, if a file is opened for Level-3 I/O, a Level-2 I/O call appears to fail.

 

In a test program I open a DF128 file and read the contents.  At EOF I close the file.

 

The test program uses a copy of Tom Bentley's "TCIO" library that I disassembled and modified.  I added two functions, 'tstat' and 'tstats'  'Tstat' uses the Level-3 STATUS (Op-code 9) command to get information about the test file.  'Tstats' uses the Level-2 Get-File-Info (0x14) command to get information.  I was going to merge the two so I can get info to load a PAB for Level-3 access.

 

Before and after each Level-3 TCIO call I use 'tstat' and 'tstats' to see how the information changes as the program performs the various library calls:

1. Use 'tstat' to get file status (Level-3 Op-code 9).  Print results and info.
2. Use 'tstats' to get file information (Level-2 Op-code 0x14)  Print results and info.
3. Do a Level-3 operation.
4. Get file status (Level-3 Op-code 9).  Print results and info. [Repeat Step 1.]
5. Get file status (Level-2 Op-code 0x14).  Print results and info. [Repeat Step 2.]

 

I found that Level-2 'Get File Info' command appears to fail with Error Code 7 (file error) if the file is opened for Level-3 access.  After closing the file the Level-2 commands once again work.

 

My merged file-info function will have to skip calling the Level-2 call if the file is open, but I do not see that being a problem.

 

Also, I hate to say that my copy of Thierry Nouspikel's TI99-Pages has the 0x14 command returning the record-count at Result-pointer+8 as a byte value.  Fred Kaal says this is a word value, and my experiments seem to indicate the record-count is a word.  Both sites have been immensely useful for my experiments.

 

K-R.

 

[+Lee Stewart]

6 hours ago, Kchula-Rrit said:

Also, I hate to say that my copy of Thierry Nouspikel's TI99-Pages has the 0x14 command returning the record-count at Result-pointer+8 as a byte value.  Fred Kaal says this is a word value, and my experiments seem to indicate the record-count is a word.

 

Yeah—the TI document, GPL Interface Specifications for the 99/4 Disk Peripheral, indeed, has it as a word. As you say, Thierry’s site is a wealth of information for our TI-99/4A, but does have errors here and there. Others and I have pointed them out over the years. Most (all?) of my discoveries could be characterized as my tripping over them. ? 

 

Thierry has it wrong for subprogram >015, as well.

 

He makes a similar error with every instance of the transfer block’s error-return location, >8350 (probably a copy/paste error), which is a byte, not a word as he indicates.

 

...lee

[Kchula-Rrit]

On 7/4/2020 at 6:29 AM, Lee Stewart said:

 

Yeah—the TI document, GPL Interface Specifications for the 99/4 Disk Peripheral, indeed, has it as a word. As you say, Thierry’s site is a wealth of information for our TI-99/4A, but does have errors here and there. Others and I have pointed them out over the years. Most (all?) of my discoveries could be characterized as my tripping over them. ? 

 

...

 

...lee

Good to know I'm not the only one that 'trips' over these things...

 

The GPL Interface Spec is probably the one document I haven't read until I'm bleary-eyed.  I figured, I'm doing assembly, not GPL, so I never thought to look at it.

 

K-R.

 

[RXB]

I have read every GPL form of Data so many times I can recite some on command.