nather@ut-emx.UUCP (Ed Nather) (02/10/89)
In article <662@mks.UUCP>, wheels@mks.UUCP (Gerry Wheeler) writes: > If I may, I'd like to make a request of you software authors out there > -- please write your programs so they will handle either type of slash > in pathnames. It would make life easier for many people. It would indeed. Most of the MS-DOS utilities will accept either one, but "subst" will not, and both MSC and Borland's linker demand backslashes. A recent conversation about Borland's new debugger, which also demands backslashes, got the helpful reply "Well, OS2 doesn't allow it, so get used to backslashes." -- Ed Nather Astronomy Dept, U of Texas @ Austin
spolsky-joel@CS.YALE.EDU (Joel Spolsky) (02/10/89)
In article <662@mks.UUCP>, wheels@mks.UUCP (Gerry Wheeler) writes: > If I may, I'd like to make a request of you software authors out there > -- please write your programs so they will handle either type of slash > in pathnames. It would make life easier for many people. Actually I noticed that if you write your program with Turbo-C it _will_ accept both kinds of slashes unless you do something to prevent it. Isn't that neat? +----------------+----------------------------------------------------------+ | Joel Spolsky | bitnet: spolsky@yalecs.bitnet uucp: ...!yale!spolsky | | | internet: spolsky@cs.yale.edu voicenet: 203-436-1483 | +----------------+----------------------------------------------------------+ #include <disclaimer.h>
sagen@nucthy.physics.orst.edu (Milt Sagen) (02/10/89)
In article <10353@ut-emx.UUCP> nather@ut-emx.UUCP (Ed Nather) writes: >A recent conversation about Borland's new debugger, which also demands >backslashes, got the helpful reply > >"Well, OS2 doesn't allow it, so get used to backslashes." > I've got to say this, it is nothing more than my opinion, so take it or leave it. Making the backslash character the directory separator was a stupid idea. Consider the fact that the backslash character is not located at any one place between keyboards. It is such a pain to go to a different machine with a different keyboard with a different location for the backslash yet running the same operating system. Milt Sagen Internet: sagen@nucthy.physics.orst.edu Department of Physics Oregon State University Corvallis, OR 97331 Tele: (503) 754-4631
kiravuo@kampi.hut.fi (Timo Kiravuo) (02/10/89)
In article <8769@orstcs.CS.ORST.EDU> sagen@nucthy.PHYSICS.ORST.EDU (Milt Sagen) writes: >Making the backslash character the directory separator was a stupid idea. It was a F***ing stupid idea. Microsoft is not very competent even at their best, but this was one of their worst ideas. And IBM has topped it. On the new 102 key Scandinavian keyboard the backslash is generated by pressing right-Alt and the key right to the 0 (+ in Scandinavia, - in the USA). Now how does the IBM think that we type backslash, when _both_ keys needed to generate it are on the right side of the keyboard. Absolutely idiotic. If they have to redesign the keyboard, why coudn't they at least have somebody to try to type on it. I mean, they have been making typewriters for such a long time, that they must have heard about touch-typing. Timo Kiravuo 90-451 4328 kiravuo@hut.fi kiravuo@fingate.bitnet opmvax::kiravuo
gbs@stratus.UUCP (George B. Smith) (02/11/89)
In article <10353@ut-emx.UUCP> nather@ut-emx.UUCP (Ed Nather) writes: >In article <662@mks.UUCP>, wheels@mks.UUCP (Gerry Wheeler) writes: >> If I may, I'd like to make a request of you software authors out there >> -- please write your programs so they will handle either type of slash >> in pathnames. It would make life easier for many people. > >It would indeed. Most of the MS-DOS utilities will accept either one, >but "subst" will not, and both MSC and Borland's linker demand backslashes. >A recent conversation about Borland's new debugger, which also demands > >"Well, OS2 doesn't allow it, so get used to backslashes." I have to whole heartedly agree with Gerry on this one. I would like to ask all software authors, commercial and public domain, please make sure your programs can handle either the forward slash or back slash character as the separator. You know it is sooOOOoooo easy to do, there is no reason that I can understand not to. Some people might say, "It is an undocumented feature so I won't use it!" Yet so much DOS software *must* use other undocumented stuff I can't see why *this* feature is singled out for avoidance. I have seen some programs published by PC Mag which uses every trick (not) in the book for TSRs but don't handle the '/' as a path separator. And it is very easy to do, as I said before. All a program has to do is find out the current setting of the switch char. This is a trivial task for a program. In fact, in Turbo C v2.0, in the file getopt.c supplied by Borland, there is the 3 lines necessary to get the current switch char. Then, if the current switch char is '-', then assume the path separator is the '/'; if the switch char is not the '-', then I would say it is safe that you can assume the path separator is the '\'. When I ported the BSD indent(1) program to DOS, I had to modify the code to handle creating backup file names. Enclosed below is the code segment for how I handled it: /* * copy input file to backup file. * * The strategy for handling appending a suffix has been modified * from the original for use on MS-DOS. * * If there is no suffix on the input file name then the suffix * ".BAK" will simply be concatenated. If there is a suffix, * then it will be stripped before the ".BAK" is concatenated. * * However, for MS-DOS, we must check to see what the *probable* * path separator character is by looking at the value of the current * command line switch character. If the 'switchar' is a '-' as in UNIX, * then we will assume the path separator is a '/'. If the 'switchar' * is not '-', then we assume it is a '/', the MS-DOS standard, and * assume the path separator character is the MS-DOS standard '\'. * * Then make the backup file the input and original input file the output. */ bakcopy() { int n, bakchn; char buff[BUFSIZ]; register char *p; char *rindex(); char path_separator; /* depends on current DOS switchar */ char suffix[4]; /* buffer for any filename suffix */ char *strsuffix(char *, char *); /* get string suffix */ /* first, construct the backup filename */ path_separator = (getswitchar() == '-') ? '/' : '\\'; if ((p = rindex(in_name, path_separator)) != NULL) p++; else p = in_name; (void)strcpy(bakfile, p); if (strsuffix(bakfile, suffix) != '\0') { p = rindex(bakfile, '.'); *p = '\0'; } (void)strcat(bakfile, ".BAK"); Again, I urge all software writers to add this very easy, nice to the user feature to their programs. It would sure make going between UNIX and DOS much more pleasant. George B. Smith Stratus Computer, Inc gbs@stratus.stratus.com {amdahl,oliveb,uunet}!stratus!gbs
alex@mks.UUCP (Alex White) (02/12/89)
In article <1623@stratus.UUCP>, gbs@stratus.UUCP (George B. Smith) writes: > In article <10353@ut-emx.UUCP> nather@ut-emx.UUCP (Ed Nather) writes: > > > >It would indeed. Most of the MS-DOS utilities will accept either one, > >but "subst" will not, and both MSC and Borland's linker demand backslashes. > >A recent conversation about Borland's new debugger, which also demands > > > >"Well, OS2 doesn't allow it, so get used to backslashes." Interestingly enough, MSC's interface ``cl'' accepts either - or / for options. It is also explicitly documented in their manuals, that / may be used in #include filenames so that you can exchange programs with Xenix. OS/2 has just made things worse - cmd.exe now takes `^' (the uparrow, cap) as an escape character, doing what `\' has always done under unix. OS/2 and DOS all continue to accept either slash in all system calls interfacing to the kernel.
vg55611@ihuxy.ATT.COM (Gopal) (02/12/89)
In article <1623@stratus.UUCP> gbs@stratus.UUCP (George B. Smith) writes: [stuff deleted] >I have to whole heartedly agree with Gerry on this one. I would like >to ask all software authors, commercial and public domain, please make >sure your programs can handle either the forward slash or back slash >character as the separator. [stuff deleted] !And it is very easy to do, as I said before. All a program has to do !is find out the current setting of the switch char. This is a trivial !task for a program. In fact, in Turbo C v2.0, in the file getopt.c !supplied by Borland, there is the 3 lines necessary to get the current !switch char. Then, if the current switch char is '-', then [stuff deleted] There is an even simpler method: allow only "-" to be used as the switchar in your programs, and then whenever you read a pathname from the user, substitute all occurences of "/" with "\". Venu P. Gopal UUCP: att!ihuxy!vg55611 Internet: vg55611@ihuxy.att.com BITNET: com%"vg55611@ihuxy.att.com" or com%"vg55611%ihuxy@research.att.com"
yogi@humus.Huji.AC.IL (Yossi Gil) (02/12/89)
According to the last rumor I heard, DOS (>= 3.0) system calls, or int 21 functions in the PC slang, treat equally the '/' and the '\'. The one to blame for the backslash backward thinking is command.com. Anyone knows more about it?
hollen@spot.megatek.uucp (Dion Hollenbeck) (02/13/89)
From article <8769@orstcs.CS.ORST.EDU>, by sagen@nucthy.physics.orst.edu (Milt Sagen): > > Making the backslash character the directory separator was a stupid idea. > Consider the fact that the backslash character is not located at any one place > between keyboards. It is such a pain to go to a different machine with a > different keyboard with a different location for the backslash yet running the > same operating system. > In addition to that it is schizophrenic for us UNIX users when switching back and forth between DOS and UNIX. Even more frustrating is using the MKS Toolkit on DOS - now two different pieces of the same computer use different switch characters! Dion Hollenbeck (619) 455-5590 x2814 Megatek Corporation, 9645 Scranton Road, San Diego, CA 92121 seismo!s3sun!megatek!hollen ames!scubed/
alanf%smile@Sun.COM (Alan Fargusson @ peace with the world) (02/14/89)
There is a small miss-conception here. MS-DOS allows both forward and backward slants in path names. It is command.com that takes the forward slant, so you cannot use it unless you change the switchchar. Programs should be coded to allow both. It is actually valid to combine both in one path name. A little history: command.com is designed to look like the CP/M command interpreter which uses the forward slant as a switchchar. Remember that MS-DOS 1.X did not have a hierarchal file system, so there was no conflict. - - - - - - - - - - - - - - - - - - - - - Alan Fargusson Sun Microsystems alanf@sun.com ..!sun!alanf
tmurphy@wpi.wpi.edu (Tom [Chris] Murphy) (02/14/89)
I once forgot where I was and used a / instead of a \ in a config.sys file under 3.2. The dammed thing worked! Until I went to 3.3 .... -- Thomas C. Murphy Worcester Polytechnic Institute CAD Lab Internet: tmurphy@zaphod.wpi.edu tmurphy@wpi.wpi.edu BITNET: TMURPHY@WPI CompuServe: 73766,130
pt@geovision.uucp (Paul Tomblin) (02/15/89)
In article <1623@stratus.UUCP> gbs@stratus.UUCP (George B. Smith) writes: >>In article <662@mks.UUCP>, wheels@mks.UUCP (Gerry Wheeler) writes: >>> If I may, I'd like to make a request of you software authors out there >>> -- please write your programs so they will handle either type of slash >>> in pathnames. It would make life easier for many people. > >I have to whole heartedly agree with Gerry on this one. I would like >to ask all software authors, commercial and public domain, please make >sure your programs can handle either the forward slash or back slash >character as the separator. I don't know if it is true of 2.0, but turboC 1.5 handled either syntax no matter what your switchar was. I was typing in the following to TurboC for a few days before I realized that I had been lucky that it had worked: tcc -I/gvc/includes -I/turboc/includes dba.c or even tcc -I/gvc/includes -I\turboc\includes c:/csource/dba.c Holey metacharacters, batman. I had tried a simple switchar program for a while, thinking it would ease my daily transition between DOS and Unix, but found that MOST commercial programs I tried, and most dos utils, don't understand it. Sigh... >...When I ported the BSD indent(1) program to DOS... Did you do this? CAN I HAVE IT?! Thanks... -- Paul Tomblin, Second Officer, Golgafrinchan B Ark | Canada's Acid Lakes: UUCP: nrcaer!cognos!geovision!pt ?? | 150,000 Points of Disclaimer: The opinions expressed here aren't | Blight. necessarily even mine! |
allbery@ncoast.ORG (Brandon S. Allbery) (02/16/89)
As quoted from <10353@ut-emx.UUCP> by nather@ut-emx.UUCP (Ed Nather):
+---------------
| "Well, OS2 doesn't allow it, so get used to backslashes."
+---------------
Who uses OS/2?
I dunno about anyone else, but I'm considering porting a simple database
manager or clone thereof (say, a clone of Informix 3.30 [pre-SQL]) to Minix
and suggesting *that* as an alternative to OS/2. It doesn't even need a 4MB
286 box to run!
++Brandon
--
Brandon S. Allbery, moderator of comp.sources.misc allbery@ncoast.org
uunet!hal.cwru.edu!ncoast!allbery ncoast!allbery@hal.cwru.edu
Send comp.sources.misc submissions to comp-sources-misc@<backbone>
NCoast Public Access UN*X - (216) 781-6201, 300/1200/2400 baud, login: makeuserzu@ethz.UUCP (Urs Zurbuchen) (02/16/89)
In article <1623@stratus.UUCP> gbs@stratus.UUCP (George B. Smith) writes: >I have to whole heartedly agree with Gerry on this one. I would like >to ask all software authors, commercial and public domain, please make >sure your programs can handle either the forward slash or back slash >character as the separator. Is there really anyone disagreeing? Life could be much more easier if it was done this way. >And it is very easy to do, as I said before. All a program has to do >is find out the current setting of the switch char. Now, I don't agree with this one. Why not allow '/' and '\' as directory separators in pathnames. In most cases you even don't have to stick with '-' as the switch character but can allow '/' as that, too. >... Then, if the current switch char is '-', then >assume the path separator is the '/'; if the switch char is not the >'-', then I would say it is safe that you can assume the path separator >is the '\'. The safest way is to test for '/' as the switch character. If so, path separator has to be '\'. Otherwise it's '/'. No need to test for '-'. Have a nice day, ...urs UUCP: ...!mcvax!cernvax!ethz!zu
gbs@stratus.UUCP (George B. Smith) (02/20/89)
In article <783@ethz.UUCP> zu@bernina.UUCP (Urs Zurbuchen) writes: >In article <1623@stratus.UUCP> gbs@stratus.UUCP (George B. Smith) writes: >>And it is very easy to do, as I said before. All a program has to do >>is find out the current setting of the switch char. >Now, I don't agree with this one. Why not allow '/' and '\' as directory >separators in pathnames. In most cases you even don't have to stick >with '-' as the switch character but can allow '/' as that, too. I would answer that I think it is ugly to have *both* backslashes *and* forward slashes in the same path name. I would say that at some level this can be considered a personal, esthetic consideration and would be very hard to agree on. I just prefer C:/top/sub1/sub2/file over C:\top/sub1\sub2/file. Although, on a technical level, mixing the separator characters would effectively disallow the use of the backslash as an escape character which is a common convention. I would hazard the guess that people who would like to use the forward slash as the path separator would like to use the backslash as the escape character. >>... if the current switch char is '-', then >>assume the path separator is the '/'; if the switch char is not the >>'-', then I would say it is safe that you can assume the path separator >>is the '\'. >The safest way is to test for '/' as the switch character. If so, path >separator has to be '\'. Otherwise it's '/'. No need to test for '-'. I agree with Urs on this point. I will change my code to do as he suggests. Thanks for the insight. > Have a nice day, > ...urs And to you. George B. Smith Stratus Computer, Inc gbs@stratus.stratus.com {amdahl,oliveb,uunet}!stratus!gbs