Comcom causing a file to be created when redirecting to device name with colon

[andrewbird]

FDPP/comcom32

C:\>echo longstring > nul:
C:\>dir
 Volume in drive c is /tmp/test-imagedir/dXXXXs/c
 Directory of c:\

2026-01-17 12:25             7 a.txt
2026-01-17 12:24            17 version.bat
2026-01-15 15:01 <DIR>         dosemu
2026-01-17 12:24            28 testit.bat
2026-01-17 12:24           622 fdppauto.bat
2026-01-17 12:24           467 fdppconf.sys
2026-01-17 12:28            13 NUL~5D
         6 file(s)           1154 bytes
         1 dir(s)            40.0 GB free

File is created test-imagedir/dXXXXs/c/nul\:

FDPP/FreeCOM

C:\>echo longstring >nul:
C:\>dir
 Volume in drive C is IR DXXXXS C
 Volume Serial Number is 4ADD-1B8A

 Directory of C:\

A        TXT             7  17/01/26 12:25
VERSION  BAT            17  17/01/26 12:24
DOSEMU               <DIR>  15/01/26 15:01
TESTIT   BAT            28  17/01/26 12:24
FDPPAUTO BAT           622  17/01/26 12:24
FDPPCONF SYS           467  17/01/26 12:24
COMMAND  COM        85.480  17/01/26 12:31
         6 file(s)         86.621 bytes
         1 dir(s)     40.977 Mega bytes free

Output is thrown away as expected.

[stsp]

Could you please check the
difference against an MFS
log? I suspect they present
the name in some different
forms, 1 of which is not
deducible for mfs.c code as
a device...

[andrewbird]

Here are the log files. The salient part seems to be around line 13007

t1.zip

It seems that FreeCOM is stripping the colon off before handing it to truename. I also note this in the FreeCOM source
https://github.com/FDOS/freecom/blob/ec6c63f13be0b254151c76cbbbec3c80ae33741b/lib/openf.c#L77-L90

[stsp]

Hmm, thanks, cool detective
work! And no need to hack
mfs.c, which is great.
Maybe you want to provide
the comcom patch before I
get to that. :)