djl@fisher.UUCP (Dan Levin N6BZA ) (09/25/85)
Found this one while trying to make sendmail work,... all the lines of
the form
From:
To:
Date:
came out of the form
Rom:
O:
Ate:
Seems that the XENIX 3.0 ctype.h has toupper() and tolower() defined as
macros like this,...
#define toupper(x) (ifsomething(x) ? dosomething(x) : (x))
which breaks badly if (x) is autoincrement or autodecrement. Ie.
toupper(s++) goes to (ifsomething(s++) ? donesomething(s++) : (s++)),
which is clearly wrong.
The fix is to ctype.h, to not use the trinary operator; or to use the
libc versions instead (by #undef'ing toupper() and tolower());
NB: This only holds for the Intel distribution, I have not checked the
IBM XENIX 3.0 release.
--
***dan
{allegra,astrovax,princeton,twg}!fisher!djl
The misplaced (You call *that* a ski slope??) Californianrlk@chinet.UUCP (Richard L. Klappal) (09/28/85)
In article <764@fisher.UUCP> djl@fisher.UUCP (Dan Levin N6BZA ) writes: >Seems that the XENIX 3.0 ctype.h has toupper() and tolower() defined as >macros like this,... >#define toupper(x) (ifsomething(x) ? dosomething(x) : (x)) >which breaks badly if (x) is autoincrement or autodecrement. Ie. >toupper(s++) goes to (ifsomething(s++) ? donesomething(s++) : (s++)), >which is clearly wrong. > >The fix is to ctype.h, to not use the trinary operator; or to use the >libc versions instead (by #undef'ing toupper() and tolower()); > >NB: This only holds for the Intel distribution, I have not checked the >IBM XENIX 3.0 release. > >-- > ***dan > >{allegra,astrovax,princeton,twg}!fisher!djl >The misplaced (You call *that* a ski slope??) Californian This is not the only bug running around in toupper/tolower. While some systems have this implementation, other versions (among which is the Fortune) do not do any sanity check, they just add/subtract 0x20, thus transforming numerals into control codes or letters, depending on the direction of conversion. Harbison & Steele warn that toupper/tolower are inconsistant between versions if *NIX. and indicate what they suggested (?? SYSV ?? ANSI ) recommendations are. My answer has been my own local ROUTINES (not macros) named 'ucase' and 'lcase' that do the sanity check and compute the result. It causes a little redundant code on machines that work right, but is nice insurance anyplace they don't, without conflicting with the internal names (viz. tolower or _tolower, etc). (Apologies if I misspelled either of the names above, it doesn't look right, but the book is upstairs) -- Richard Klappal UUCP: ..!ihnp4!chinet!uklpl!rlk | "Money is truthful. If a man MCIMail: rklappal | speaks of his honor, make him Compuserve: 74106,1021 | pay cash." USPS: 1 S 299 Danby Street | Villa Park IL 60181 | Lazarus Long TEL: (312) 620-4988 | (aka R. Heinlein) -------------------------------------------------------------------------
tg@sfmin.UUCP (T.Glinos) (09/29/85)
> Found this one while trying to make sendmail work,... all the lines of > the form > From: > To: > Date: > came out of the form > Rom: > O: > Ate: > > Seems that the XENIX 3.0 ctype.h has toupper() and tolower() defined as > macros like this,... > #define toupper(x) (ifsomething(x) ? dosomething(x) : (x)) > which breaks badly if (x) is autoincrement or autodecrement. Ie. > toupper(s++) goes to (ifsomething(s++) ? donesomething(s++) : (s++)), > which is clearly wrong. > > The fix is to ctype.h, to not use the trinary operator; or to use the > libc versions instead (by #undef'ing toupper() and tolower()); > > NB: This only holds for the Intel distribution, I have not checked the > IBM XENIX 3.0 release. > > -- > ***dan > > {allegra,astrovax,princeton,twg}!fisher!djl > The misplaced (You call *that* a ski slope??) Californian *** REPLACE THIS LINE WITH YOUR MESSAGE *** XENIX System V has this all fixed.
henry@utzoo.UUCP (Henry Spencer) (10/01/85)
> This is not the only bug running around in toupper/tolower. While > some systems have this implementation, other versions (among which > is the Fortune) do not do any sanity check, they just add/subtract > 0x20, thus transforming numerals into control codes or letters, > depending on the direction of conversion... This was the original definition of toupper/tolower. It was not a bug; the domain of legal arguments was letters of the opposite case, period. This way if you knew the nature of the argument, the overhead of the check was unnecessary; if you didn't, you were supposed to do the check first. So many people were sloppy about it that inadequate domain checking became a characteristic error when using these macros. So the definition was eventually changed. Implementations that don't check are not buggy: they are conforming correctly to an earlier, and arguably less desirable (certainly less idiot-proof), version of the specs. -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry
addw@root44.UUCP (Alain Williams) (10/02/85)
>> Seems that the XENIX 3.0 ctype.h has toupper() and tolower() defined as >> macros like this,... >> #define toupper(x) (ifsomething(x) ? dosomething(x) : (x)) >> which breaks badly if (x) is autoincrement or autodecrement. Ie. >> toupper(s++) goes to (ifsomething(s++) ? donesomething(s++) : (s++)), >> which is clearly wrong. This is clearly right. When you are dealing with macros, you must always expect this. The fact that it broke news software is a reflection on the way news was written, not the XENIX macros. -- Alain Williams, Root Computers Ltd, London, England. <UK-net>!{hrc63|qmc-cs|edai|kcl-cs|ukc|glasgow|ist|jmccfr|west44 ubu|rlvd|pmllab|stc}!root44!addw {unisoft|deccra}!root44!addw root44!{rootcl|rootis}!addw