En Thu, 15 Feb 2007 11:34:53 -0300, John Machin <[EMAIL PROTECTED]> escribió:
>> > Probably not. AUX, CON, NUL, PRN, LPT, COM1, COM2 etc (with or without >> > an extension) are reserved in Windows for specific devices for >> > compatibility with MS-DOS 1.00 programs, which did that for >> > compatibility with CP/M. >> >> (This is OT now) Do you know why "AUX.csv" is invalid too? I can accept >> that AUX (without extension) is an invalid filename, but it is quite >> different from "AUX.csv" >> > > It is actually a valid file name, but the file is not on disk. I > presume that the OP got an error because it was in 'a' (append) mode > which requires an existing disk file. See below. > > C:\junk>copy con aux.csv > fubar > ^Z > 1 file(s) copied. The above gives me an "Access denied" error; perhaps because AUX is my serial port and it is currently in use. > Why? Who knows? We're talking CP/M, MS-DOS and Windows and you want to > know why? Probably too lazy to distinguish between 'AUX\0', 'AUX.\0' > and 'AUX.XYZ\0' ... probably stopped scanning on reaching the first > invalid character. If you're desperate to find out, dial up your > nearest RCPM and ask the sysop :-) Ahhhh... I think you hit the point, indirectly. On CP/M the filename was not stored as 'AUX\0' - remember, ONLY 8 characters plus 3 for extension, and NO PATH. A FileControlBlock (FCB) had exactly 11 characters reserved for the file name (plus the drive number). So it was actually "AUX " vs "AUX XYZ" and... well, the lazyness argument again. -- Gabriel Genellina -- http://mail.python.org/mailman/listinfo/python-list
