Mycology

Mycology is my Befunge–98 test suite. The ‘Fungus’ Funge–98 diagnostics tool was inadequate for my purposes, with inaccurate output and very little coverage. I developed Mycology side by side with my CCBI interpreter, to the point that it now tests the full Befunge–98 specification, with the only exception of the = Execute instruction, which is too difficult to reliably test across platforms.

Mycology is written completely in Befunge–98.

A comparison of Mycology test results across interpreters is available. In summary: CCBI passes with flying colours, most others don’t. *grin* But I must say that the situation has improved very well since I released Mycology.

Downloads

The latest version, 2009–04–04:

.txt readme, 22.3 KiB
Source:
- .7z, 42.7 KiB
- .zip, 47.0 KiB

Significant changes

Not all minor updates and details are included here. To see a full list of changes, view the readme.

2009–03–29

Properly fixed the 1y bit testing, it was quite broken.

2009–03–28

R in FILE is now expected to reflect when reading past EOF.
STRN doesn’t test G leading to an infinite loop, since I think it’s valid that it does loop infinitely.
It was always claimed that I/O was buffered; fixed that.
It was complained if 1y pushed something greater than 15: corrected that to 31.

2008–11–15

) was never actually tested with a negative count, it was ( both times.

2008–09–21

The fingerprint loading code had a misalignment which caused a broken ( to loop infinitely; fixed that.

2008–09–13

“GOOD: SGML spaces” was never output.
HRTI doesn’t now run ffkyn, filling up the stack so much at once.
IMAP updated to latest spec.
INDV now expects the logical order, reporting BAD otherwise.
Fixed: test for k with a negative count was expecting pre–2008–07–26 k behaviour.

2008–09–10

The check for wraparound with a non-cardinal delta is a bit stricter.

2008–08–30

The case when SCKE is included in SOCK now works as well.

2008–08–28

Fixed: many of the SOCK and SCKE tests still (since the 2008–07–26 version) expected that A overwrites its input socket.

2008–08–20

Fixed: the concurrency test for ;; was off by one.
Added testing for the new A and O in SUBR.

2008–08–19

Added the DATE fingerprint.
Fixed: the concurrency tests for 5kz and "a b" were incorrect.

2008–08–13

I had got the way y works as a pick instruction completely wrong. (Reading the spec, it seems pretty clear… I wonder how I managed that.)
PNTR is gone, after INDV it shouldn’t have existed any more.
D in FILE is now tested, and used to remove those pesky .tmp files.

2008–07–26

It turns out that k is meant to behave differently from what I thought. Nothing k-related is UNDEF any more, thanks to information from Chris Pressey (via Arvid Norlander and Mike Riley).
This also meant some changes in MycoTRDS.
Added a k test related to the new behaviour.
Expanded the null byte check somewhat.
Updated the readme for the first time since the first release! I completely forgot about it…
In SOCK, overwriting a socket on accept is now BAD, not GOOD.

2008–07–19

Added a check for null byte handling.

2008–03–11

i and o being unimplemented or unable to be tested are now UNDEF, not BAD.
PERL fingerprint is now tested with the string “5–1”, not “2+2” which is invariant under reversal. Thanks to Alex Smith for the input.

The comparison hasn’t been updated to reflect these changes.

2008–02–02

The behaviour of 1k # is now considered UNDEF.