[comp.lang.c] C compilers: Opinions solicited

karthur@codas.att.com (Kurt_R_Arthur) (12/24/87)

Since the release of Microsoft C version 5 and Turbo C 1.5, I am going to retire
my old C compiler (upgrade is not a possibility) and purchaseone of the
following:

	1. Microsoft version 5

	2. Borland Turbo C version 1.5

	3. Datalight Optimum C (Version unknown).

Considerations:

	1. Price  approximate MAIL ORDER prices: 
		Turbo 		$ 57 (source for libraries $150)
		Microsoft	 265 (is source included/avail?)
		Datalight	  95 (is source included/avail?)

	2. MS-DOS Performance - faster the better (of course!)  Compilation
		speed is of (next to) no importance, but execution speed
		is imperative!

	3. OS/2 compatibility - I want to be able to write applications for
		all flavors of OS/2 (standard, extended and the 
		Presentation Manager using dynalinks, etc.) without buying
		another compiler.

	4. Libraries: small, medium, large, huge, graphics, etc.

	5. Debug environment is nice, but not necessary.

	6. Integrated environments without a command line interface will
		immediately eliminate a compiler from the competition (I
		have an editor I like already).


Experiences, comments, statistics, bugs, better prices (if known) are
appreciated.  If I get enough data I will post them to the net.

Thanks in advance,

Kurt Arthur
Software Services of Florida.

manes@dasys1.UUCP (Steve Manes) (12/25/87)

In article <1455@codas.att.com> karthur@codas.att.com (Kurt_R_Arthur) writes:
>Since the release of Microsoft C version 5 and Turbo C 1.5, I am going to retire
>my old C compiler and purchase one of the following:
>
>	1. Microsoft version 5
>	2. Borland Turbo C version 1.5
>	3. Datalight Optimum C (Version unknown).
>Experiences, comments, statistics, bugs, better prices (if known) are
>appreciated.

I don't know the Datalight compiler nor do I know how library-compatible
it is with the others.  That's an important consideration if you use any
canned commercial or PD C libraries.  MSC and Turbo are the two most
supported with Lattice following closely.

I've got both Turbo and the full MSC 5.0.  I really like Turbo's editor and
its general speed by comparison to MSC.  I also like the in-line assembler
(I know, I know... gross nonstandardization but if portability is an issue
you don't use assembler code anyway).  However, in the first 30 minutes I
discovered that Turbo doesn't like my EGA (a Mitsuba) and anything compiled
with Turbo would crash my machine.  This includes a pre-compiled MicroEMACS
I pulled off some d/l board that I later discovered was compiled using
Turbo.  A friend also had similar trouble on a different EGA board.  That
made Turbo unusable for me since my main development machine has an EGA and
many of my clients do too.  This problem makes Turbo a poor choice
for outside-world development.

MSC provided the most seamless change of environment for me.  It also
includes a wealth of library functions that MSC has needed for a long time
(maybe MS took a cue from Lattice here).  The compiler seems to run a
little faster than 4.0 as do the executables however the execs are a little
fat compared to MSC 4.0.  The only trouble I had with the conversion was
#undefing the toupper and tolower macros if ctype.h is loaded.  Calls like
toupper( getchar() ) will not work properly if the macros are used.  In
fact, I've been unable to get the toupper/lower macros to work with any
function as an argument.

Quick-C is nice for quick 'n dirty programming, like writing a cr/lf
converter, but a pain for anything larger.  I also think it's inferior to
Turbo's editor, ergonomically speaking.

-- 
+-----------------------------------------------------------------------
+ Steve Manes         Roxy Recorders, Inc.                 NYC
+ decvax!philabs!cmcl2!hombre!magpie!manes       Magpie BBS: 212-420-0527
+ uunet!iuvax!bsu-cs!zoo-hq!magpie!manes              300/1200/2400

scott@hpcvcg.HP (Scott Linn) (12/30/87)

Just another possibility...

I see that MIX is now selling Power C, a new C compiler (presumably
different than MIX C, which was very slow).  The price is 19.95, and
they will even sell the source (I think the source is around $20).

Anyway, it seems *very* cheap, and worth a try.  Their new ad was in
the latest Dr. Dobbs (or Computer Language).  Power C has at least three
memory models that I know of.  I bought MIX C a while back, and have
been generally pleased, although it is slow.  Power C looks to be *much*
faster.

Anyway, I would look into it.

Scott Linn
Hewlett-Packard Northwest IC Disivion