The chart of comp.lang.c

[Chris Hills]

It's not really about direct control, though that's part of it.

If you have an IDE that can share projects between three (or more!)
platforms and whose editor can be persuaded to act like vi, I want to
know about it

When you say cross platform do you mean target or host? Most will run
on Windows but few will run on other hosts. Though people inthe
embeded world tend to write on one host fro many targets.

Several companies will supply an IDE with their compilers that will give
you a single IDE that can at the click of a mouse target any one of 30
targets.

Most editors in IDE's will do the vi bindings.

If in fact "most" developers prefer not to use an IDE (a claim I believe
to be true, but not one I'm prepared to defend),

I think most use the IDE but like you not something I want to get in to
long arguments about. ( I am a tool distributor and I talk to 100's of
developers a year about tools also I meet lots of them at conferences
etc)

the biggest reason is
probably that most IDEs have weaknesses in areas that are important to
them that their favorite non-IDE tools don't have.

Not so these days. Also most IDEs automate many areas of a project.

(yes, I do work on three different platforms, and have at least one
hobby project that will eventually "need" to run on all three)

Fair enough.

Then you are probably not going to find an IDE for your work BTW which
3 platforms.

 

[CBFalconer]

.... snip ...


I think most use the IDE but like you not something I want to get
in to long arguments about. ( I am a tool distributor and I talk
to 100's of developers a year about tools also I meet lots of them
at conferences etc)

IMO the principle use for an IDE is to avoid learning the
intricacies and interactions of a group of development tools.
Those same tools may often be much more useful when separately
controlled.

 

[Chris Hills]

IMO the principle use for an IDE is to avoid learning the
intricacies and interactions of a group of development tools.

Yes and often saving much time.

Those same tools may often be much more useful when separately
controlled.

Sometimes but not usually. For the majority IDE's save a lot of time
and trouble.

It is very rare when dealing with my customers for a wide range of
compilers I ever have to go "under the hood" on an IDE. This is for a
lot of people doing embedded software, usually on custom hardware.

On the whole the IDE is a fast and effective way of working.

Some people said much the same when manual advance/retard and chokes
were removed on cars.... would you advocate going back to having manual
advance/retard on cars?

The old ALWAYS say the new is not as good as the old days. You may be
happy and comfortable with separate tools and spending time and effort
getting them to work together but I find the IDE's much better (and I
have used command line, separate tools and make for many years)

 

[Jean-Marc Bourguet]

IMO the principle use for an IDE is to avoid learning the
intricacies and interactions of a group of development tools.
Those same tools may often be much more useful when separately
controlled.

Most people don't learn the intricacies and the interactions of the group
of development tools they use. A small part of the development team set it
up and the other use their set up.

An IDE don't change that. A small part of the development team setup the
group of development tools (included the IDE) and the other use that set
up. For simple project, the IDE can simplify that work. For more
complicated one, it is an additional burden.

The IDE then provide some help than some find usefull -- nearly all people
having learn with IDEs, and some other -- and that I've never seen on non
integrated environement (mostly very context sensitive help). I don't have
an opinion about the real usefullness of that because I've never used an
IDE on a real world project... I've tried several time but the one I'm
working on are big enough and old enough to have been set up without
thinking about IDE. And making any IDE works with an existing setup seems
to be a major work I'm not ready to take on just to check if it is better
than what I currently have (a manually built environment on top of emacs
which works on all platform we support) and I've currently saw no potential
benefit of using the IDE without a proper setup.

Yours,

 

[jacob navia]

Most people don't learn the intricacies and the interactions of the group
of development tools they use. A small part of the development team set it
up and the other use their set up.

An IDE don't change that. A small part of the development team setup the
group of development tools (included the IDE) and the other use that set
up. For simple project, the IDE can simplify that work. For more
complicated one, it is an additional burden.

The IDE then provide some help than some find usefull -- nearly all people
having learn with IDEs, and some other -- and that I've never seen on non
integrated environement (mostly very context sensitive help). I don't have
an opinion about the real usefullness of that because I've never used an
IDE on a real world project... I've tried several time but the one I'm
working on are big enough and old enough to have been set up without
thinking about IDE. And making any IDE works with an existing setup seems
to be a major work I'm not ready to take on just to check if it is better
than what I currently have (a manually built environment on top of emacs
which works on all platform we support) and I've currently saw no potential
benefit of using the IDE without a proper setup.

Yours,

The problem is that under linux there is no IDE. Linux people say "IDEs
are not useful" because they never saw one. All "IDE"s under linux are
just front ends to gdb/gcc without any added value. There is seldom "go
to definition", integrated search, project management, etc etc. Mostly
the IDEs under linux are bad quality, maybe with the exception of
IBM's Eclipse.

Why?

Because nobody wants to pay for them under linux. So they get stuck
with vi/gdb/gcc, the development tools I always use when I have to
work under that system.

Vi/make for editing/compiling
vi+gdb for debugging.

Cumbersome, and very horrible, but it is the customer that decides.

 

[Chris Hills]

The problem is that under linux there is no IDE. Linux people say "IDEs
are not useful" because they never saw one.

I wondered why there was this anti IDE philosophy around. It was
similar when people developed under Unix.

All "IDE"s under linux are
just front ends to gdb/gcc without any added value.

That explains a lot.

There is seldom "go to definition", integrated search, project
management, etc etc. Mostly
the IDEs under linux are bad quality, maybe with the exception of
IBM's Eclipse.

Why?

Because nobody wants to pay for them under linux.

This is a major problem. Until people get used ot paying for SW under
linux a lot of the Sw won't appear on Linux.

IBM put money behind Eclipse

 

[CBFalconer]

.... snip ...

The problem is that under linux there is no IDE. Linux people say
"IDEs are not useful" because they never saw one. All "IDE"s under
linux are just front ends to gdb/gcc without any added value.
There is seldom "go to definition", integrated search, project
management, etc etc. Mostly the IDEs under linux are bad quality,
maybe with the exception of IBM's Eclipse.

Nonsense. All IDEs are really supervising other discrete
programs. At most the editor can be specialized. Many Linux users
are sufficiently curious that they understand the basic programs,
and can dispense with the IDE.

 

[Al Balmer]

Fair enough.

Then you are probably not going to find an IDE for your work BTW which
3 platforms.

There are multiplatform tools which may not call themselves IDE's, but
perform most of the functions of an IDE. I'm thinking particularly of
Slickedit, which I use on Windows, HP-UX, and Linux. Also, isn't
Eclipse usable on a number of platforms?

 

[Jean-Marc Bourguet]

Jean-Marc Bourguet wrote:

The problem is that under linux there is no IDE. Linux people say "IDEs
are not useful" because they never saw one.

Could you name an IDE? I know people who don't find Visual Studio useful.
One of them specially was happy to switch to a project under windows so
that he could use it and he is now using emacs under windows and VS only as
a front-end for the debugger and he is complaining that his environment is
more difficult to use with VS compiler that with mingw which use gdb...

My position is that I've yet to find an IDE which I could make a use of in
my context in the limited time I've had to make them work. So I can't
speak about their usefulness.

I don't say it is impossible to make them usable in my context. They all
have their notion of how a project should be set up. And I can't change
the way our 30000+ files application is setup just to test one IDE on one
platform. And I was unable to addapt the IDE I've tried to our project.

All "IDE"s under linux are just front ends to gdb/gcc without any added
value. There is seldom "go to definition", integrated search, project
management, etc etc. Mostly the IDEs under linux are bad quality, maybe
with the exception of IBM's Eclipse.

For a whole week, I've tried to use Eclipse without success -- not full
time, I've a job to do. It seem to want to parse all our files, and crash
out of memory after a long time in our case. Desactivating that, you have
something no better than emacs and I even couldn't make it usable as a
front end for debugger (for example, doing a step took 10s or so, from the
gdb log, I guessed trying to update panes I had hidden in an attempt to
speed up things).

Because nobody wants to pay for them under linux. So they get stuck with
vi/gdb/gcc, the development tools I always use when I have to work under
that system.

Vi/make for editing/compiling
vi+gdb for debugging.

We'd be ready to pay for something more productive than our current setup.

Currently, I've reverted to emacs+gdb from inside emacs+my macros. More
cumbersome than an IDE properly setup? Perhaps, I've yet to be able to try
a properly setup IDE. For sure there is too much work in my macros.

Yours,

 

[Richard Heathfield]

Chris Hills said:

I wondered why there was this anti IDE philosophy around. It was
similar when people developed under Unix.

Linux /is/ an IDE. Anyway, the reasoning that Linux people say IDEs are
not useful because they never saw one is specious. I'm a classic
counter-example. I'm a "Linux person", I suppose, since I use Linux for
programming, for reading and sending email, for reading and sending
news, for graphics, for animation, and even for the occasional game.
Nevertheless, I've used IDE (the Turbo C IDE), Borland's Win32 C and
C++ Builder IDEs, RHIDE (an IDE clone shipped with DJGPP), PWB (anyone
remember that?), and Visual Studio (all versions since Version 1 with
the possible exception of one or two of the later versions).

Although a Linux person, I am not "anti-IDE". As I say, I consider Linux
to *be* an IDE (albeit an extraordinarily powerful and expressive one).
And on those occasions when I find myself using Windows, I will
cheerfully use an IDE.

So it is simply not true that I am anti-IDE. What I am anti is non-C
subjects being discussed in comp.lang.c.

<snip>

 

[Chris Hills]

Chris Hills said:


Linux /is/ an IDE.

No Linux is an operating system.

Anyway, the reasoning that Linux people say IDEs are
not useful because they never saw one is specious. I'm a classic
counter-example. I'm a "Linux person", I suppose, since I use Linux for
programming, for reading and sending email, for reading and sending
news, for graphics, for animation, and even for the occasional game.

Are you confusing a GUI with an IDE?

 

[Chris Hills]

Nonsense. All IDEs are really supervising other discrete
programs. At most the editor can be specialized. Many Linux users
are sufficiently curious that they understand the basic programs,
and can dispense with the IDE.

What does curiosity have to do with it.

I used to use Make a lot and discrete programs but I more often use an
IDE these days for speed.

Curiosity has nothing to do with it.

Of course if your time counts for nothing and you are happy to keep
re-inventing the wheel that is up to you.

 

[Chris Hills]

Could you name an IDE?

IAR EWB
Multi Edit
uvison
Code Warrior
There are lots of them If I could be bothered to check the
names/spellings I can give you over 20.

It depends on your environment and what you want to do. Personally I can
get projects up and running in the IDE's I have far faster than trying
to knit my own system together.

 

[Kenny McCormack]

No Linux is an operating system.

No, it's a desert topping.

 

[Richard Heathfield]

Chris Hills said:

No Linux is an operating system.

That, too. But it's still an IDE. It is the environment in which I
develop, and its various parts are integrated. It is, therefore, an
integrated development environment, as far as at least one person
(myself) is concerned.

Are you confusing a GUI with an IDE?

No, you have selectively quoted me. In the above part of the paragraph,
I was merely presenting my reasons for considering myself included
within the term "Linux person". I invite you to read the rest of that
paragraph, within the original article, before deciding whether or not
you believe that I know what an IDE is.

 

[Jean-Marc Bourguet]

Note: I've set the Follow-Up set to poster.

IAR EWB
Multi Edit
uvison
Code Warrior

I've had a quick look at Multi-Edit and didn't see any reason to switch to
it from my customized emacs set up.

There are lots of them If I could be bothered to check the names/spellings
I can give you over 20.

It depends on your environment

We currently support Linux, Solaris, AIX, HP-UX in 32 and 64 bits. My
development platform switched from Sparc/Solaris to x86/Linux recently.
And will perhaps switch back to x86/Solaris in the future.

A somewhat big existing project (35000 files in several languages, multiple
teams around the world, several streams of support). Quite a big inertia
also.

and what you want to do.

- Immediate goal: Have a better productivity...

I feel limited by the tools I'm using and I'm looking for something better.
As a matter of fact, I'd be happy with a correct GUI front-end for gdb.
I've downloaded and tested several things and I always reverted back to gdb
launched inside emacs.

So the IDEs don't compete with "build something now" instead, but with
something already built and working. So it has to warrant migration cost
and can't disturb too much thoose needing or wanting to stay in the current
environment.

- More generally: know what if anything can an IDE bring to me now or later
in another set up in my current set up if too peculiar. And for that I
need to be able to test them in a real life situation.


My major gripe on IDE currently is that I can't test them as they are too
big a pain to adapt to something existing (I never succeeded) and using
simple approaches leads to performance problems like the "45+ minutes then
crash" start up or the 10s for a next inside the debugger of Eclipse I've
already written about.

The nearest thing I've used to an IDE is Sun's Workshop (it is was I used
on Sparc/Solaris), and it is not supported for several years now. I've
tried several times SunStudio, but each time it was just too slow to be
usable.

Personally I can get projects up and running in the IDE's I have far
faster than trying to knit my own system together.

That's not a problem I've. Things are already set up for the kind of
programs I work on professionnally and for personal project I've used my
own build system taking for years -- starting a new project is typing

start_project project_name

and that's all.

 

[Richard]

Chris Hills said:


That, too. But it's still an IDE. It is the environment in which I
develop, and its various parts are integrated. It is, therefore, an
integrated development environment, as far as at least one person
(myself) is concerned.

No. Linux is an operating system. You are confusing it with the Gnu
development tools.

No, you have selectively quoted me. In the above part of the paragraph,
I was merely presenting my reasons for considering myself included
within the term "Linux person". I invite you to read the rest of that
paragraph, within the original article, before deciding whether or not
you believe that I know what an IDE is.

The only IDE worth considering (gnu tools on the command line is not an
IDE because they are not "integrated") is eclipse or ... emacs.

--

 

[jacob navia]

The only IDE worth considering (gnu tools on the command line is not an
IDE because they are not "integrated") is eclipse or ... emacs.

For instance (just a small example) a good IDE does
EDIT+CONTINUE.

You find a bug in the code, correct it, and the IDE compiles
in the background, incorporates the code into the running
program, and goes on debugging!

This is science fiction for linux!

I have tried to do it but it is too much work. The IDE of
lcc-win32 doesn't have this feature.

 

[CBFalconer]

IAR EWB
Multi Edit
uvison
Code Warrior
There are lots of them If I could be bothered to check the
names/spellings I can give you over 20.

It depends on your environment and what you want to do.
Personally I can get projects up and running in the IDE's I have
far faster than trying to knit my own system together.

Oh? For example, under Windoze, I have a console window and an
editor window open. The editor window can access (and list) all
files for a project. The editing sequence is:

ALT-TAB
<edit>
CTL-S (save)
ALT-TAB
make
run (or go back to step 1)
all done (or go back to step 1)

very quick, very flexible, and the error listings are sitting right
there in the command line window when needed. If I wish I can
open further windows for manuals, to examine data files, whatever.
I don't have to change the mechanisms under Linux.

Conceded - operation on an independent embedded system complicates.

 

[Ian Collins]

For instance (just a small example) a good IDE does
EDIT+CONTINUE.

You don't require an IDE for that, a decent debugger should do the job.
Or does this make the debugger an IDE?