Borland C++ Now Free-as-in-Beer

BlueBlade writes, "Inprise, formerly known as Borland (and now merging with Corel) has released their latest compiler for free. They said the move was to show support for the open source community. You can find the full article here. " It's Free Beer though. No source code, although such a release would really show their support for the Open Source Community.

Comments

jeeze...

[PimpBot]

Its Free Beer tho. No source code, although such a release would really show their support for the Open Source Community.

Ever heard of being thankful for what you get? Borland makes quality stuff, and its great to see this come out. But, CmdrTaco, your comment sounds (at least to me) resentful. Lighten up a bit, ok?
--------------------------

Re:jeeze...

[zorgon]

Agreed! C'mon, Rob, it's a compiler and libraries. Not everyone needs to work under the hood on that part of it all. Go forth and build cool open source stuff with it, said the voice of the meta-Philippe to the teeming masses. That's pretty darn supportive if ya ask me. I think "free beer" is a bit too derogatory, as every schoolchild knows you only rent or borrow beer and have to give most of it back in short order -- too much like a shrinkwrap license to me! Z.

"C makes it easy to shoot yourself in the foot; C++ makes it harder, but when you do it blows your whole leg off."

Re:jeeze...

[Zibblsnrt]

Well, then all the zealots would be whining anyway.. "What are stories about windows-only release doing on slashdot?!!" You just can't win anymore.

And it supports open source in pretty much the same way handing an artist some good, free paints supports art. Who cares if the compiler doesn't have the source? It's what you make with it that matters.

-Patrick Stewart

Re:jeeze...

[Signail11]

I'm sorry, but you obviously haven't compared the quality of the code generated by gcc with that generated by any Borland, Microsoft, or Intel's C compiler. Frankly, the commercial compilers blow gcc out of the water, even with pgcc -O6 -march=pentiumpro -fomit-frame-pointer -etc. gcc is *extremely* impressive for what it does, as it is portable to a zillion different OSes and CPUs, but this very portability makes it less than optimal at generating code for any one CPU. The internal RTL representation does not lend itself to the detailed modelling of the Intel x86 crappy archetecture so critical to acheiving good performance, especially with a stack-based FPU.

Undue Pushing...

[Deus Ex Machina]

Alright, I know I'm probably going to get moderated down for this one, but here goes... Why is it that whenever a company releases a powerful and useful tool for free, the first response of many here is to say "Great, now give us the source and we will like you!" I'm not trying to rant or make any personal griping here, but to me it is in very bad form to expect so much of a company when they are still in a transition toward the entire concept of Open Source Software. I do not believe that Any Company, when approached with the concept of releasing their tools for free, looks upon the idea without a jaundiced eye at first, and I belive that it is only after weighing their options and realizing the goal that they make an action like the one Inprise made today. It is a big deal to them, and they are being good sports and catering to our community with something like this... so why must we grab their outstreched hand and tug them toward us with all our might? Ah well... it just seems rude to me. Moderate away.

Re:Undue Pushing...

[GregWebb]

This only works if they don't consider it better than the rival products, I'm afraid.

They're not directly making money off this any more, but it's still used (according to the press release) as the base compiler for C++ Builder. Which they at least try and make money from, though I haven't seen the balance sheets and can't say whether they actually do :)

If they don't think it's anything special, they might consider opening the source under a particular license - but why would people work on it rather than on GCC?

On the other hand if they consider it special then it's a potential selling point for C++ Builder. But if they open the source then they're leaving the oppportunity for a rival product to use their code. Philosophically nice but it could hurt their sales if (for example) Visual C++ could now use each and every trick that C++ Builder's compiler does to produce nice code as well as their own, unknown tricks. At which point they've done something which reduces the profitability of the company and that isn't legal for a publically traded company in the UK so I can't see the US investors looking that kindly on it.

I can see why you want it, I just don't think it's all that realistic.

Greg

Once upon a time...

[jd]

Borland's biggest claim to fame was writing the best Pascal compiler on the planet. After that, they wrote Pascal 4.0 and dropped off the face of the planet.

Now, they're releasing the binaries for zero cost. It's a start, but not really a very helpful one. Yes, they seem to have learned that "Free Software" is a strength, not a weakness, but they seem to be confusing free with "free beer", not "free speech".

IMHO, this is a regrettable confusion. A blunder. Nothing to flame them over, but rather something to gently correct them for. Borland has some great brains, and if they can be tapped to their full potential, both Borland AND the Free Software community will gain.

However, zero-pricing their binaries doesn't really benefit anyone. It takes commercial companies away from genuinely Free Software, thus dividing the potential for growth, rather than multiplying it.

Still, there IS the possibility that this is but one move in a steady transition to genuine Free Software, be it BSD-ish, GPL-ish, etc. It might be best to wait a little and see.

Anyone look at system requirements?

[theSheep]

It's Windows only, guys. And without the source, it's impossible for anyone outside Borland to port it. At least for myself, this is pretty much useless, and I suspect a lot of the /. crowd does mostly Linux development.

I wouldn't be surprised to see a port at some point, though given that it's owned by Corel.

Re:Anyone look at system requirements?

[Ronin Developer]

A port of BCC 5.5 to linux already exists. This was demonstrated at the 1999 Borland/Inprise Conference. Yes, the version released is for Windows. But, they demonstrated that they could run the same source through their Linux compiler, have it compile, and run as expected.

Borland is planning to release Kylix which includes (to my understanding) Delphi on Linux as well as C++ Builder for Linux.

If anyone has used their tools, you will find that they provide all the source code to the libraries (mainly the VCL). They don't include the source to the compilers or other tools (they need to make money somehow, right? Isn't that what for-profit corporations do?).

Nobody is forcing you to download the BCC. Nor, are they forcing you to visit their site. The fact that you went there says you are interested.

The only reasons I can think that they want to know who is downloading their compiler (i.e registration) and filling in the survey are:

1) Knowing if their marketing campaign is having an effect.

2) Being able to notify registered users when Kylix becomes available.

In a previous post, somebody griped about the BCC compiler stating that the Cygnus tools (even MS's own were preferable to BCC). The only downside I ever experienced using BCC was that many libraries in the Windows world simply would not link to it. But, I experienced the same problem with MS's and Symmantec's compilers as well. At least in the Linux world, there is a standard on how these libraries should be built. The Borland compilers should be compatible with other libraries on Linux. No?

So, why does everyone keep knocking Inprise/Borland for their approach of giving away the product?

Personally, it seems to me that too many people don't want to look at the good that can come from this. They expect everything to be free or a threat to their purist, unrealistic, view of the world. That's great for hobbiests. It doesn't make money. Or, perhaps they are staunch MS supporters who don't want to see Inprise/Borland succeed? Just read their posts and decide for yourself.

Is it compatible with Visual C/C++?

[SIGFPE]

Microsoft's compiler is the least ANSI compliant compiler I know and I can barely use the STL with it. Could I use the Borland compiler as a drop in replacement? For example I need to write DLL's to be loaded up by code compiled with Visual C/C++ and that need to link against libraries compiled with Visual C/C++. Is there any chance that I could use Borland's compiler to do this?

I've never understood how developers manage under Windows when Microsoft's compiler doesn't actually compile C++

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:Still no match for the GCC

[jd]

What compile flags were you using?

To get decent speed, you need -at least- a -O2, preferably a -O6, though not all code will work correctly at that level and the optimiser can be shaky at times.

You really also need to turn off any and all debug flags, strip the binaries of symbol tables, and link to libraries which are compiled the same way as above.

(If you've a heavily optimised program linked to a library that's space-compacted, with debug flags up to it's eyebrows, you're really not going to get any decent speed out of it, no matter how good the compiler is.)

Lastly, watch those -m flags. If you aren't using maths, don't use them. -m486 will slow your code down, horribly, because of the increased amount of data it passes in each function and procedure call.

Yes, the GNU compiler -COULD- be improved, enormously. I think that a decent multi-pass compiler, with intelligent flag control, would be great. A multi-pass linker would be cool, too - I am fed up of errors due to putting library calls in the "wrong" order. It's quite capable of doing a once-through to search for symbols, and working out how to link from there.

Personally, the best compiler I've ever seen was the Oxford Computer Systems Petspeed compiler, which was a 4-pass compiler/linker. That thing generated decent code.

Comment removed

[account_deleted]

Comment removed based on user account deletion

All the "Whining": My 2c

[Caspian]

I'd just like to state that the person who pointed out how Inprise/Borland has taken great pains to portray themselves as the benefactors of the "Open Source" community had a VERY good point. What is going on here is essentially the cheapening of what free software stands for. It used to be that people would use the term "free software"-- but that is all about freedom, which conjures up all sorts of unpleasant (at least, to Corporate America's eyes) imagery of scruffy revolutionaries sitting in smoky bars sipping cognac, and so the term "open source" was adopted by ESR, thereby dissociating the code from any immediately obvious social issues and turning it purely into a technical or economic issue. Alright, but then what? Then, all the companies that released even the slightest bit of information, or the smallest of binaries, started flinging about the word "Open"; now, here comes Borland (or whoever the hell owns them now) trying to paint themselves as being in the same boat as the "Open Source" community, while meanwhile all it is doing is giving away (after a mandatory registration AND a survey, I might add) a very much closed-source compiler, for Windows only (!!!), with no debugger?!

I fear that the logical continuation of this pattern is as follows-- since Borland has just established that by giving away free-beer-free goodies (not even necessarily whole software packages-- arguably, a development package is not complete without a debugger!) somehow puts one in the same boat as the Open Source/free software crowd, the next steps involve doing the same for (in order):

* Crippleware/nagware/software that times out after X days-- "But see, it's helping them until it expires! They can use it to create Open Source applications for 30 days!"

* Inexpensive software ("But see, it's cheap.. we're not greedy, so we're helping the Open Source community!")

And, eventually, anyone selling their piece of software for a few bucks less than their competitors will attempt to brand themselves as being pillars of the Open Source community...

But again, that's just my 2c.

Re:All the "Whining": My 2c

[nazerim]

* Crippleware/nagware/software that times out after X days-- "But see, it's helping them until it expires! They can use it to create Open Source applications for 30 days!"
* Inexpensive software ("But see, it's cheap.. we're not greedy, so we're helping the Open Source community!")

It's called shareware. And the press release doesn't say anything about being "the pillars of the open source community". They simply stated that a free ANSI C++ compiler is now available for the Open Source community to do as they wish. Nothing more, nothing less.

You might scoff at this, but note that there *is* no other freely available native compiler for ANSI C++ available for the Win32 platform. That's the significance of Borland's announcement. ANSI C++ is a very difficult animal to tame. Why are we all moaning? C'mon, cut them some slack, they're already making a Linux version of their software, no reason why C++ on their Linux beast won't compile on their Win32 toy.

We'll all reap the benefits, damnit.

Re:Can somebody explain...

[HeUnique]

OK, I'll answer this..

When you grab this compiler, you can use it, but you don't have to. You can use either MS, GNU tools, and lots of others..

With the IE example - you HAVE to install it. Try for example to install Visual Studio without any version of Internet Explorer. You can't. period. Why? because MS decided that in order to use Visual studio you'll have to use THEIR browser. Why can't it install only the essential parts (online help reader etc..) and the other functions will run on another browser? because MS doesn't want to.

Now do u understand why people like this news about BC++ and NOT IE?

reverse moderator psychology

[CoughDropAddict]

While we're griping, let me tell you about something that I'M sick of: whining about how sure you are you'll get moderated down, as a cheap shot at evoking the sympathy of the moderators.

From now on, whenever I'm a moderator, I'll specifically look for people who think they can buy a few karma points by making themselves look like martyrs for their opinions and I will moderate them down, like they supposedly expect to be anyway.

If you have an insightful comment to make, the least you can do is let it get moderated up by virtue of its thoughtfulness, without having to resort to begging.

Product abandonment

[Animats]

Is this an end-of-life "make it free and abandon it" move? That's becoming common. Sun did that with their Java Workstation product (their rather lame attempt at an IDE for Java). It's a reasonable move, but not worth publicizing.

Probally because of IP

[dieman]

I bet they can't release the compiler because of IP issues within the compiler. That seems to be the number one reason of legacy software not being able to be released to public.

Malda: Not everyone has to go opensource. I think you need a little action with the clue-by-four.

gcc doesn't "include a debugger"

[CentrX]

Does that make it "ABSOLUTELY WORTHLESS"? No. You can use debuggers that are part of a separate package, such as gdb.

Also, I beg of people NOT to e-mail anyone at Borland to complain about this. It is a good gesture on their part to release the compiler free, there's no need to complain that they didn't release everything free.

Chris Hagar

Actually Borland is quite superior to GCC

[bbcat]

>With GCC you get a lot more optimizations and a
>lot more active development of the product.

What have you been sniffing?

I did some comparisions with my program compiled
with GCC and with Borland C++ 5.5

I had normal optimization with Borland and
highest optimization with GCC. I also compiled
with Visual Age for C++ with normal Optimization.

Size with GCC : 2.82M
Size with Borland : 1.18M
Size with IBM : 1.25M

The code with Borland is also slightly faster
than that of IBM and GCC code, GCC having
the slowest code and Borland the fastest.

My only complaint now that I figured out which
flags to use is that it doesn't support long long.

>With Borland 5.5 you only have Corel/Inprise
>developing it. Borland claims their 5.5 compiles
>code faster than GCC and their product does have
>precompiled headers.

All true. It is fast, it will be great to use
for Linux if it is just as good as the winblows
version. It'll be great when the C++ builder
and Delphi comes out as well.

Hands up everyone who doesn't get open source

[khiron]

Borland releasing a free version of the compiler does NOT mean that the compiler is Open Source. What it means numbskulls is that if you as a customer buy an open source product that is built using C Builder, and want to compile it, you do not need to buy the compiler. IF you want to make substantial changes then no doubt you will WANT to get all the source to the VCL which is after all 90% of a builder app and you will be glad to supports Borlands efforts by buying it. Borland's language products have ALWAYS had a strong following because they ship the source to most of their libraries (instead of enmiring you in evil activex, et al), heck TP3 shipped the source code for a spreadsheet as sample code. Borland was open source long before it was fashionable. Expect the compiler for Pascal to be released in the very near future (it is the same backend), and the Linux compilers (Kylix) to follow. The advantage for Open Source developers is that you can now release your shource freely knowing that there will allways be a refference compiler that is fully ANSI compliant, works equally with Windows and Linux (We've all got our fingers crossed that this will be true) and is fast. And if nobody buys Borlands products they dissapear.

Some comments on Borland & Open Source

[Dacta]

Firstly, they aren't Inprise any more - they are dropping that name (at last!) in the merger with Corel. Infact, you could actually buy the Inprise sign on Ebay - it might still be there.

Secondly, Borland is considering open-sourceing more of its stuff. They are already releasing Interbase under the MPL (see www.interbase.com) and in a recent interview in the Linux Journal, Dale Fuller said they were considering what they shoudl open source with the release of Kylix (Delphi/C++ Builder for Linux). This MAY EVEN INCLUDE THE VCL. That woudl be really, really cool.

Don't Care, just want compiler

[kevin805]

I don't really care whether it's open source. I really want a compiler for windows. I'm not about to pay $1000 for visual studio, though. So I'm really happy about this. Socialists can go whine elsewhere. You want a compiler, write it. Borland is offering something. If you don't like it, fine. If it benefits some people, fine. If it conflicts with your political views, too bad. You ever hear "don't look a gift horse in the mouth"?

For the truely paranoid, though, Corel now has a way to make a proprietary Linux distribution. They still have to open the source to their mods, but it doesn't do you much good if it will only compile with Borland's proprietary compiler.

Link libs != DLL's

[Venomous Louse]

The link libs are incompatible. That's what you use implib for. It generates Borland-friendly link libs from arbitrary DLL's. MSVC doesn't have an analagous utility (BTW MSVC 5 is violently unable to cope with MSVC 6 link libs if they've got a $debug section, ha ha -- found that out the hard way, installed MSVC 6, and wasted a day and a half fixing the damage to my system from the "upgrade"). 32-bit windows DLL's are goddamn PE files, they all have the same header and the same sections and blah blah blah. You're talking out your ass.

If you don't like the struct alignment, change it. Compilers can do that.

For anybody in the audience not familiar with dynamic linking in Win32: The library is called foobar.dll, and it comes with a smaller file called foobar.lib, which is what you "link" to, and which contains basically the names of the exported functions and not much else. The compiler groks all this.

Kinda Off Topic Re:jeeze...

[extrasolar]

Not everyone needs to work under the hood on that part of it all.

This is why the Free Software and Open Source philosphies are so different. The Open Source philosophy, being mostly pragmatic, has the unstated opinion that just because the user probably doesn't need or want the source code, it is okay to be taken away. The Free Software philosophy beleives freedom should be granted for all its users, whether or not it is pragmatically beneficial.

Note that a philosophy is different from a community. I actually don't beleive that there are distinct Free Software and Open Source communities, per se. People who don't share in *part* of the Free Software philosophy are probably those who haven't heard of it. I mean, come on, there are no big highlines on ZDNET or CNN about the Free Software Revolution lately, are there? It seems like Linux and Open Source have stolen the hype, but that's okay. And I don't think there are any serious Free Software people who say "I actually prefer crappy software."

The free compiler from Borland represents a misunderstanding about Open Source, the other half, Free Software. I think Open Source and Free Software are two halfs of the same entity. Once they figure out that the community isn't after purely pragmatic goals, then maybe they will change the license to something more free?

Note: I am not attacking you or anyone else. Please don't be hostile in your reply.

Well they kinda release source

[guran]

Borland does release the source for the class libraries together with their C++ builder and Delphi tools. Really nice actually. I use Delphi a lot.

Those are not free(beer) though. And the source for the compilers is still closed....