12 Jun
2018
12 Jun
'18
5:53 p.m.
On 2018-06-10 02:27:27 David C. Rankin wrote:
On 06/09/2018 10:16 PM, Leslie Turriff wrote:
Since my
background is in the green-screen environment of IBM's mainframe
computers, I prefer editors that don't require me to stop typing and use the
mouse to perform certain operations, and I'm used to having actual commands
(as opposed to ctlrl or alt sequences. The editor I'm using today is Blair
Thompson's X2, but it has some irritating bugs in the Unix version that
haven't been addressed. The worst problem I have with it is that I can't
override the font size in the X Windows version, and I need a microscope to
read its text. Since Blair doesn't seem to be interested in fixing these
minor issues, I've been looking for what I consider a suitable replacement
for X2.
I'm using the source code for the
open-source Thompson-Davis Editor
<http://adoxa.altervista.org/tde/>, "a simple, public domain, multi-file,
multi-window binary and text file editor written for IBM PCs and close
compatibles running DOS, Win32 (console) or Linux" to learn C and curses
programming.
You may have noticed its coincidental acronym :-) and yes, it stores
it's configuration information in $TDEHOME. TDE is small but
non-trivial, and I think it will be a good resource for someone like me,
who is attempting to get a grip on programming in the *nix environment;
it's not so big as to be overwhelming, yet not as useless as
HelloWorld.c.
Thoughts?
Leslie
Leslie,
If you truly want to learn C, and you are looking for editors, you have
two basic classes of editors you can use (both more than capable). Your
choices are either traditional text-console/curses based (e.g. vim, emacs
and what sounds like Thompson-Davis is), or you have your ui-toolkit based
editors like kate, kwrite, geany (https://github.com/geany/geany) or
codeblocks (http://www.codeblocks.org/downloads).
Of the two, I prefer the ui-toolkit based editors, but I'm at home with
vim as well. In the ui category -- you really cannot do better than
kate/kwrite (and if you aren't smart enough to use TDE, then a geany or
codeblocks work well).
For the text/curses based class, it's a preferences thing. vim, or emacs
are both incredibly capable, though I find emacs a bit heavier than vim.
Learning C is one of the best choices you can make. Nothing else (aside
from assembly) provides you the complete control over hardware as C does.
However, working at the hardware level comes with great responsibility. You
are responsible for memory management. One-byte too few and you walk off
into the realm of Undefined Behavior. Leaning C, you learn to program and
will be a better programmer as a result, regardless of which language you
ultimately end up writing in.
I tend to agree. I started in the computer industry
in the early '70s (long
before PCs were thought of), so I know how to program, but not in C; my
familiarity is with FORTRAN, COBOL and Rexx. Finding useful examples of C
code that I can use to get a comprehensive feel for the language is
difficult; Hello World is so trivial as to be useless, and finding something
that actually does useful work while being small enough to get my head around
has been difficult. Hopefully, with TDEditor I'll be able to kill two birds
with one stone, learning to write C and producing a text editor that has the
features I want.
The best advice anyone can give you in learning C is simply to slow-down.
C is an exact language. Understand every declaration, the proper use of
each library function and for God sake validate the return of every
allocation and every I/O call. You don't digest the whole library at once.
When you use a library function, open the man page -- read it -- pay
careful attention to all parameter type, return types and the RETURN
section of the page.
C has a certain reputation as a language that makes it very
easy for a
neophyte programmer to shoot himself in the foot; I intend to be very
careful. :-)
(pay particular attention to the man page for the scanf family -- which
provides no end of pitfall for the new C programmer who fails to learn the
distinction between a *matching* and *input* failure and fails to
understand that when either occur, no further characters are extracted from
the input stream and are left unread -- just waiting to bite you on your
next call)
When you are learning, forget an IDE, build your code on the command line
so you actually learn what the compiler option are and how to correctly use
them -- then you can properly use an IDE. Until you understand how to
compile and link your code -- there is no way you can tell an IDE how to do
it.
When I learned to program there was no such thing as an IDE, and I've
never
found them particularly useful. (When I started out, my IDE was an IBM 029
keypunch machine. The closest I got to an IDE on the mainframe was ISPF's
PDF editor's ability to insert code snippets. :-) )
*Always* compile with *warnings enabled* and do not accept code until it
compile cleanly, without warning. For gcc/clang that means adding -Wall
-Wextra -pedantic (and I would add -Wshadow) to your command line. (clang
also provides -Weverything). If you hop on a windoze box, the for VS
(cl.exe) use /W3 (/Wall provides a great deal of non-code specific warnings
that are a bit much).
Okay. I must note, though, that many of the packages that
I've installed
from source tarballs are liberally sprinkled with warnings, even though their
authors presumably consider their code to be production quality.
In TDEditor, for example, gcc complains about these two code snippets:
| if (token == "User" ||
| (parse_token( line, token ), stricmp( token, "User" )) == 0)
in config.c produces
| Compiling config.c
| config.c: In function ‘process_menu’:
| config.c:1988:23: warning: comparison with string literal results in
unspecified behavior [-Waddress]
| if (token == "User" ||
| ^
which makes some sense, though in the languages I'm familiar with such a
construction's behaviour is well-defined; I would guess that defining "User"
as a string constant would make the compiler happy?
and
| static const char *fc, *fc2;
in pull.c produces
| Compiling pull.c
| pull.c: In function ‘make_menu’:
| pull.c:835:25: warning: variable ‘fc2’ set but not used
[-Wunused-but-set-variable]
| static const char *fc, *fc2;
| ^
which makes no sense at all, since
| fc2 = (frame < 3) ? fc : graphic_char[2];
certainly looks to me like it's being used. I can see where C got that
reputation. :-)
Finally, go get an account on StackOverflow.com -- seriously. If you have
a C question -- it has probably already be answered 50 times over, and the
answer peer-reviewed. (the same applies for just about any language, it is
a fantastic site) Before you post a question you are expected to have
searched for a similar one and taken the http://stackoverflow.com/tour,
visited http://stackoverflow.com/questions/how-to-ask and when posting
always provide a http://stackoverflow.com/help/mcve (a minimal, complete
and verifiable example -- meaning something those there can compile to
verify your code behavior and help debug)
I've used StackOverflow many times,
though I don't have an account there.
What's the advantage of having an account?
Learning C isn't something you'll do in a day, or a week, or month, or
year -- it is a journey -- enjoy the ride.
No kidding. :-)
Leslie