diff options
Diffstat (limited to 'whichbas.rst')
| -rw-r--r-- | whichbas.rst | 175 |
1 files changed, 148 insertions, 27 deletions
diff --git a/whichbas.rst b/whichbas.rst index 5e7363c..8a7e15e 100644 --- a/whichbas.rst +++ b/whichbas.rst @@ -10,19 +10,58 @@ Determine BASIC variant of a tokenized Atari 8-bit program SYNOPSIS ======== -whichbas [-v] *input-file* +whichbas [-v] *input-file* [*input-file* ...] DESCRIPTION =========== -**whichbas** reads a tokenized Atari 8-bit BASIC, Turbo BASIC, -BASIC XL, BASIC XE, or Atari Microsoft BASIC program and attempts to -discover which BASIC is required to run it. +**whichbas** reads tokenized Atari 8-bit BASIC, Turbo BASIC, +BASIC XL, BASIC XE, BASIC/A+, or Atari Microsoft BASIC programs and attempts to +discover which BASIC is required to run each one. + +*input-file*\s must be actual files. **whichbas** can't read from +standard input, because it seeks in the input file. + +With multiple *input-file*\s, the output is one line per file, +with the filename, a tab, and the detected BASIC. With only one +*input-file*, the filename and tab are not printed. + +OPTIONS +======= + +*Note*\: none of the options **-s** **-k** **-v** are allowed +when checking multiple files; they only work if there's just one +*input-file* argument. + +Detection Options +----------------- + +**-s** + Script (or silent) mode. Instead of printing a human-readable + name like "Turbo BASIC XL" or "OSS BASIC XE" to stdout, **whichbas** + will print nothing on standard output, but will exit with a status + indicating what it detected. The caller can check the return status + (e.g. the **$?** variable in Bourne/POSIX shells, or **ERRORLEVEL** + in MS-DOS or Windows). See **EXIT STATUS**, below. + +**-k** + Keep going. The default is to stop looking at the program if the + BASIC type gets narrowed down to either Turbo BASIC or BASIC XE. + This option also enables **-v** (verbose). It's really only useful + for testing, if you're hacking on **whichbas** itself. + +.. include:: genopts.rst NOTES ===== +Atari BASIC programs are detected *almost* 100% reliably. See **BUGS**, below, +for the gory details. + Turbo BASIC, BASIC XL, and BASIC XE are all supersets of Atari BASIC. If you wrote a program using one of them, but didn't use any of the extra commands or functions, the result is still an Atari BASIC program. +Likewise, if you wrote a program using BASIC XE, but only used the +commands/functions it has in common with BASIC XL, it would be +detected as BASIC XL. There are two types of BASIC XE programs: regular and *EXTEND*\ed. The extended type is detected 100% reliably, because the first byte of the @@ -30,46 +69,128 @@ file changes from **$00** to **$DD**. Non-extended programs are only identified as BASIC XE if they use any of the extra commands BASIC XE adds to those found in BASIC XL. -Atari BASIC programs can be detected 100% reliably. - Detection of Turbo vs. BXL/BXE isn't 100% reliable, and probably never will be. There's too much overlap between the sets of extra tokens added by each. Programs that don't use very many of the extra -functions provided by Turbo/BXL/BXE may show up as "Not Atari BASIC; -probably either Turbo or BXL/BXE". +functions provided by Turbo/BXL/BXE may show up as "Either Turbo BASIC XL +or OSS BASIC XL". + +Atari Microsoft BASIC is detected by checking that: -Atari Microsoft BASIC is detected by checking that the first two -bytes of the file are not zero, and that the last 3 are zero. This -may result in false positives (files that aren't BASIC programs at + - the first byte of the file is **$00**\. + + - the next 2 bytes (LSB/MSB) are the same as the actual file length minus 3. + + - the last 3 bytes of the file are zero. + +This may result in false positives (files that aren't BASIC programs at all might show up as Microsoft). Also, no distinction is made between Atari MS BASIC 1.0 and 2.0. +OSS BASIC/A+ is an extended form of Atari BASIC. It's source +compatible with Atari BASIC, but not token-compatible because it uses +different token numbers for the regular BASIC commands and operators. +Example: SAVE is token **$19** in Atari BASIC (also Turbo/BXL/BXE), +but in A+ it's token **$1D**. Detection should be 100% reliable, but +since there aren't many BASIC/A+ programs in the wild, it hasn't been +thoroughly tested. + Various non-BASIC files are detected (including Mac/65 source, ELF binaries, etc) as a convenience, but I wouldn't rely on **whichbas**\'s non-BASIC file type detection if I were you. +If you need the **file**\(1) command, use it. + +**whichbas** knows nothing about other BASICs such as Frost BASIC or +Altirra BASIC. + +When using multiple *input-file* arguments, a separate instance of +**whichbas** is spawned for each file. This isn't the most efficient +approach, but it shouldn't cause problems on reasonably modern +systems. + +BUGS +==== +Misdetection +------------ +It's possible to get a BASIC XL/XE program to misdetect as Atari BASIC +by writing a program that does these things: + + - Dimension a string array with a number of elements that isn't just a + numeric constant or numeric variable (e.g. *DIM A$(2+2,10)* or + *DIM A$(I*2,10)*). The code that detects a string array *DIM* command + can only handle simple cases like *DIM A$(10,10)* or *DIM A$(I,10)*. + It doesn't actually matter if the 2nd argument is a fancy expression, + though. + + - Does *not* actually *use* the string array variable by assigning to + it or reading a value from it. String array accesses in BASIC XL/XE + are reliably detected because they require a semicolon after the + first number, even if there isn't a 2nd number. Example: *? A$(2;)* + prints the 2nd element of the *A$* string array. If it were written + as *? A$(2)*, but *A$* is a string array (not a regular string), + BASIC XL/XE would give an *Error 40* (string type mismatch) at + runtime. + +The good news is, such a program will still work fine in Atari BASIC. +Atari BASIC will dimension it as a regular string variable and ignore +the 2nd dimension. Since it's not actually used elsewhere in the program, +it doesn't hurt anything. -LIMITATIONS +EXIT STATUS =========== -Currently, **whichbas** doesn't look at the variable name or type -tables. One problem caused by this: If a program uses only Atari BASIC -tokens, but uses variable(s) with _ in the name, it will be identified -as Atari BASIC... even though _ in variable names is illegal in Atari -BASIC and pretty much guarantees the program is Turbo/BXL/BXE. -Looking at the variable types could also improve detection, since -Turbo and BXL/BXE support extended variable types. +Without the **-s** option, exit status is 0 for success, non-zero for +failure. -**whichbas** knows nothing about other BASICs such as Frost BASIC, -BASIC/A+, Altirra BASIC... +With the **-s** option, the exit status is: -OPTIONS -======= +**0** + Not used with **-s**. -.. include:: genopts.rst +**1** or **2** + Error reading file. Error message will be printed on standard error. -EXIT STATUS -=========== +**3** + Atari BASIC detected. + +**4** + Turbo BASIC detected. + +**5** + OSS BASIC XL detected. + +**6** + Non-EXTENDed OSS BASIC XE detected. + +**7** + Turbo BASIC or BASIC XL (undecided which). + +**8** + Turbo BASIC or non-EXTENDed BASIC XE (undecided which). + +**9** + Turbo BASIC, BASIC XL, or non-EXTENDed BASIC XE (undecided which). + +**10** + Unknown Atari BASIC derivative (not Atari BASIC, but not + Turbo/BXL/BXE/A+ either). + +**11** + Atari Microsoft BASIC detected. + +**12** + EXTENDed OSS BASIC XE detected. + +**13** + Compiled Turbo BASIC detected. + +**14** + OSS BASIC/A+ detected. + +**64** + None of the above; not BASIC. -0 for success, 1 for failure. +In the future, more exit codes may be defined (in the range **15** to +**63**), but the existing ones will not change. .. include:: manftr.rst |