Ports System As A Strategy Against .NET?

proclus writes: "The FreeBSD ports system has been ported to Mac OSX, GNU/Linux, LinuxPPC, and OpenBSD. Check out this descriptive paper and roll your own ports-based distribution." Besides an some informative description of the mechanics of the port system, the paper lays out the case for ports (free and readily available) as a good antidote for .Net and other subscription-based systems.

Re:Simple breakdown

It does, although IIRC building dependencies for the source isn't really a stable feature yet. From the man page for apt-get:

source causes apt-get to fetch source packages. APT will examine the available packages to decide which source package to fetch. It will then find and download into the current directory the newest available version of that source package. Source packages are tracked separately from binary packages via deb-src type lines in the sources.list(5) file. This probably will mean that you will not get the same source as the package you have installed or as you could install. If the --compile options is specified then the package will be compiled to a binary .deb using dpkg-buildpackage, if --download- only is specified then the source package will not be unpacked.

A specific source version can be retrieved by postfixing the source name with an equals and then the version to fetch, similar to the mechanism used for the package files. This enables exact matching of the source package name and version, implicitly enabling the APT::Get::Only-Source option.

Note that source packages are not tracked like binary packages, they exist only in the current directory and are similar to downloading source tar balls.

However, that will only download/build/install the one source package. If one wants to install the packages required to satisfy the build dependencies for a given source package, the build-dep option should be used with apt-get.

Re:I don't get the .NET connection...

What does this have to do with .NET; I assumed there would be some kind of intermediate code base or something. But all I see is an open platform.

I suggest that "c" is functioning as the intermediate language. Following me on this one.

When is .Net bytecode going to get compiled down for applications? Such large applications are suppose to be converted to native during the install. When are these BSD packages convered to native? Well, one can do it when one installs a program (one could also grab a pre-compiled binary for one's specific system if available, but one doesn't have to).

What does distributing in the source code gain? The ability to check for security problems and to compile down to one's specific systems. What does distributing in the byte code get? The ability to check for security (by having a logical sandbox) and compile down to one's specific system.

So I suggest the use of .net's bytecode is really rather analogous to the free software communities' use of source code. So in practice, this set of highly portable software depending on parts of FreeBSD as a "runtime" is a system that is offering many of the same benfits as the .net platform.

I suspect Timothy was meaning some simular.

Ben

I don't get it

I don't get this article at all. It's full of jargon, hyperbole and muddlethink. I'm sure there's a good idea buried in here somewhere, but I can't find it.

These ports have tens of thousands of interconnections, called dependencies, which must be satisfied in order to build the applications.

No port has tens of thousands of dependencies. And since you only build one port at a time, there is no need to take into account the dependencies of any unrelated port.

Such a large and complex network of software dependencies...

It's not a complex network. It's a simple tree.

an uber-system has been superposed on the ports system

Time to go back to English class!

...the naive user will not have to face a daunting tangle of dependency.

A) Give your users respect. They are not naive. B) Dependencies are not tangled.

Thus, the FreeBSD ports system, now as a cross-platform, globally distributed, cooperative development and distribution system could form a nexus of user freedom and empowerment.

It's nice to read that on the 4th of July, but what the fsck does it mean?!?

p.s. Why is he calling this version of Darwin "GNU-Darwin"? Is this a GNU project? Does he think that Darwin is really the GNU System?

Re:Why separate porting sources trees are evil

Not at all! I said that every developer is different. By all means, do not do what I do!

However, I can offer one solution. If you are releasing net-snmp for use on FreeBSD, I can only assume that you test it on FreeBSD, and thus have a FreeBSD box somewhere to use. Why not become the port maintainer yourself?

I don't get it. Explain?

This is going to sound trollish, so let me apologize in advance.

I guess I'm not up-to-snuff in my BSDisms. What is this ports facility, and what does it have to do with .NET? I read the article and was even more confused. I see it has something to do with package dependancies, but that's about it.

Can someone spare me a clue?

Why separate porting sources trees are evil

There are serious problems with porting trees that need to be addressed by their maintainers (but frequently aren't). Specifically, patches that go into ports trees or are applied to binary releases are rarely given back to the developers of the original package. I've been distributing the net-snmp (was ucd-snmp) [net-snmp.org] toolkit for years now and nothing infuriates me more than running across a patch I haven't seen before being applied to some distributions private code cache.

By far, the worst distributor in this regcard is RedHat. I figured for years they were merely building the source and distributing the binaries as is. The first time I looked at the spec file I found around 4 patches I had never seen before and didn't even know some of the bugs existed. Well, I thought, I'll contact them and see whats up. "Sorry, I'll try to make sure patches get your way in the future". Of course, when I checked the source rpm for the next release I found yet more patches I'd never seen... Sigh... I don't have a solution to this problem, though an obvious one might be something along the lines of at least asking the maintainer if they wished to receive CVS messages or CVS patches on a regular basis from the ports trees, or to be added as a default contact for the external bug database for a given package (To solve the redhat problem, I'm forced to go search through their bug database occasionally). Don't get me wrong, I think distributing the source and binaries in an external "easy-to-use" fashion is a great thing. What I consider wrong is to not at least mention to the original developers that changes have been made. Sure, its legal and even goes along with the licence in most cases, but in the long run I would think it would save the ports maintainers a lot of conflict merging if they kept in touch with the package developers.

Re:Why separate porting sources trees are evil

• Case in point: net-snmp on FreeBSD-4.3. The majority of patches are to makefiles. The majority of the rest only updated the FreeBSD version macros. Actual code changes are rare, and are only for the purpose of getting the software to build.

But see, you missed the point. We try to release the net-snmp software for use on FreeBSD so why wouldn't I want to see even those makefile patches so I can try to solve your problems in the main distribution? What if people don't use the ports tree and download the real release? What if they want the most recent CVS code from our cvs server?

Following your example, should I go remove all the freebsd specific code from the net-snmp distribution because the proper place for it is in the ports tree since no one else but freebsd folk will need it?

Re:Why MS and .NET will win

All the more power to them, however the community should focus on creating, and making things better, not trying to pick fights

Hallowe'en Documents.

"Linux is like Communism"

"The GPL will steal all of your hard-earned IP."

"The GPL is like Pac-man."

"We're going to support FreeBSD but not Linux because the license is better...for developers. Really."

Who's the one picking a fight here?

...hasn't anyone ever thought that there are Windows programmers who develop things on their own, post them at sites like Tucows, and are actually happy with using Windows.

Oh, I know there are people perfectly happy with using and developing on Windows. I don't wish to deny them that choice. The problem is, Microsoft wishes to deny me the choice to use anything else, by making sure Microsoft "standards" are more prevalent than any other "standards", real or perceived, and ensuring you can only take advantage of MS "standards" on MS platforms. Individual actions alone may not be "smoking guns," but the sum of their actions and behaviour towards any potential competitors and developers leads me to believe they wish to deny me and millions of others a choice we don't begrudge their customers.

That is wrong. I don't mind them innovating. I do mind them assimilating and trying to make sure the only way is the Microsoft way. I'm a consumer too, and I demand a choice of software and available tools, even if MS wishes to deny me one.

Call it paranoia, but that's the view from here.

Re:Why not wait and see what .NET really is

Every time Dr. Evil announces that he is working on a new weapon, you folks always assume it is
going to be some orbiting death-ray, a volcano machine, or a bomb that will blow the earth into
little bits. You are always jumping to conclusions...

Maybe this time Dr. Evil is making a weapon that will fight crime and make our streets safe. Or
one that only works against crooked lawyers and politicians. But I repeat myself. Did you even
consider the possibility that Dr. Evil might be trying to be helpful this time? No, you didn't.

Frankly, if I were Dr. Evil, I would be pretty upset with this constant stereotyping. Maybe that
is the cause of his inner anger that causes him to do these things. You are to blame, not him...

Re:Yumm... Ports!

I used to agree with you, until I got into Debian. Apt truly rules! This isn't partisan (well, okay, it is, but it's for very good, objective reasons). The Debian Apt system is really, honestly, the best software-getter I've ever seen. In conjunction with the Debian Packaging Guidelines, which cover many very important rules for the creation of Debian packages (basically, that it should follow certain well-defined procedures regarding libraries, configuration requirements, file locations etc), Apt really spoils the sysadmin. It would be very painful for me to move to anything other than Debian now...

Why MS and .NET will win

What possible significance could .NET have in such a world, where thousands of free software applications can be readily downloaded and configured especially for you, especially for a computer that is optimized according to your own personal needs and desires and none other? This
is the world where the user operates the distribution building tools, and we now have all of the components at hand, which are required to make this world real.

It's instances like this which will push MS over Unix in the end. " we now have all of the components at hand, which are required to make this world real. " For business that have been using MS based products for years, many have made money using Windows so why would they want to switch when people keep up with the name calling and finger pointing? (re: GPL arguments vs. MS and vice versa)

This week, Microsoft announced that it will work with Corel to port the .NET Common Language Infrastructure and the C# programming language to open-source OS FreeBSD, a Linux competitor. Microsoft submitted the Common Language Infrastructure and C# to the ECMA standards body last October, and the company says that the FreeBSD implementation will be the first on a platform other than Windows. The company believes these tools will be used for academic, research, debugging, and learning purposes on FreeBSD.

For a company so evil, at least they're extending a hand, but according to some this is viewed as MS looking to stir up troubles in the open source community. Maybe so, but how is this comment any different from stirring up the same type of bias "What possible significance could .NET have in such a world, where thousands of free software applications can be readily downloaded and configured especially for you" hasn't anyone ever thought that there are Windows programmers who develop things on their own, post them at sites like Tucows, and are actually happy with using Windows.

All the more power to them, however the community should focus on creating, and making things better, not trying to pick fights. I used FreeBSD at home and Open for my server, and have a laptop with W2K that hasn't been used in eons, and each serve their own purpose, bottom line. Comments and write ups so biased to little to sway my vote of confidence in any OS just because someone claims it to be so much better. No sirs I'll be the judge of that as will most others, so why waste time beating a dead horse. It's these same comments used against the open source community.

Everyone wants to jump in on the action, and post why they're better, and oh by the way here are 30,000 more free programs. Yes 30,000 more free programs, 30,000 more comments, and now the whole concept is lost isn't it. Meanwhile MS stands out because they focus. So please focus on making things better not worse with such biasedness

I love the BSD Ports...

There's nothing like reading about an application out there and wondering if its in the ports collection.

I had never tried GAIM, and wasn't sure it was available for BSD. But wow.. there it was in the ports...

make
make install

Groovy. Now I have GAIM. I can really dig this whole PORTS thing. :)

"Everything you know is wrong. (And stupid.)"

Re:Yumm... Ports!

One of the nice things of the ports system is that it installs all add-ons in /usr/local, not in the standard /usr/lib, /usr/bin and /bin directories. I hate my bare standard operating system being filled with garbage of add-on or optional products. No UNIX does that, except for Linux.

Further it encourages source instead of binary based distribution, still makes using the sources just as easy as using binary distributions for other package systems. The advantages of that are:

• Less chance for virii or trojan horses
• Often smaller files to be sent
• Updates to the package/port itself, not to the original application, only need tiny diffs to be sent.

Another plus (which it shared with debians apt) is automatic dependancies. A port that needs other ports/libraries will automatically get and install those as needed, you don't need to think about prerequisites as with rpm.

Re:Simple breakdown

What's wrong with grabbing a source package and doing rpm --rebuild? Okay, that's two steps instead of one, but it's not that much of a big deal.

Well, first you have to FIND the package. Then you can download it. Then build it.

Ports and apt-get (Debian) use standard distro archives. You never need to FIND the package. This may seem like a little step improvement, but it leads to a HUGE increase in user-friendliness.

As for apt vs ports - both are great and both work well. Competition is good. One can expect each of them to pull other free unices up to their level in the next two years. And that is good for everyone.

Simple breakdown

Ports: suck source and dependent source down across the net, configure for your system, build, install.

Apt-get for Debian: suck binaries down across the net, resolve dependencies, install

All other distros: trying to catch up.

Ports is even a step more fine grained than apt-get, simply because it works with source, and incompatibilities are nearly impossible (the package will refuse to build instead).

Which article did you read?

I suspect that you didn't read the article at all. Your comments have nothing to do with the article which in turn had nothing to do with .NET. Here's the quote from the article that mentions .NET.

Thus, the FreeBSD ports system, now as a cross-platform, globally distributed, cooperative development and distribution system could form a nexus of user freedom and empowerment. What possible significance could .NET have in such a world, where thousands of free software applications can be readily downloaded and configured especially for you, especially for a computer that is optimized according to your own personal needs and desires and none other?

This somehow implies that being able to quickly download and Open Source applications is somehow in competition with .NET which is about XML web services [xml.com]. It is a thing of particular bemusement to me that Open Source advocates and Slashdot editors keep attacking a .NET which is a figment of their imaginations and has nothing to do with what truly constitutes .NET (which can be gleaned from just reading the .NET website [microsoft.com]).

On second thought there is one way one might consider that this competes with .NET. The vision of .NET doing all sorts of RPC with XML over HTTP as the protocol to access web services (e.g. obtaining the current headlines on slashdot, stock quotes, perform a translation, or some other interesting web service [xmethods.com]), the author of the original article may have been trying to say that having access to the multitude of Open Source applications out there makes web services redundant since you could just download an Open Source app to do what the web service does. This is probably true for a subset of web services but things like a realtime flight tracker provided by the airline's website, UPS's package tracker, real time stock quotes, or other information that belongs to a company that you cannot directly access is where web services will shine and simply downloading Open Source apps or various screen scrapers won't cut it.

--

Re:I don't get it. Explain?

I think I get what the writers of the article were getting at.

Why use .NET (which lets people not worry about having the newest software installed on their system) when you can just use something like ports and keep the software on your system?

Windows users are sick of .DLL hell and having to upgrade software because of bugs and such. .NET will let them keep all logic off their computer and only keep the "fluffy" presentation layer stored locally. It's a good idea.

But not very practical -- with ports (and Debian's apt) it's easy to keep everything up to date and in perfect working order without any real clue as to what's going on.

hi

Hello. Could one of you PLEASEPLEASEPLEASE explain to me what this is and how it goes against Microsoft's .NET? Thank you.

If you get a chance, could you explain:

-how does it make it easier for systems to communicate with each other? (soap stuff)
-how does this make it easier for people working in multiple langauges to integrate their stuff? (you know, people in vb inheriting classes from c++, people in delphi using source written in c#, etc)
-how does this make it easier to write ASP applications that don't care what browser you're using? (webforms)

Major thanks in advance

Peace,
Amit
ICQ 77863057

Re:Why MS and .NET will NOT win

First off, MS never 'extends a hand' to anyone. They are merely offering a subset of the service so as to appear 'cross platform'.

Just like everything they do, it's all about 'embrace, extend, extinguish'. Whatever bones they throw to the free side, it will always lag behind, and be feature poor, compared to their flagship.

This attempt to take over the server side authentication process may succeed, but it may fail...just because companies don't want to be a slave to MS.

It's impossible to make the argument that MS has ever done anything to 'help' their users...but they continue to try and find ways to extract as much money as possible from their user's pocketbooks. Server side authentication and proprietary services will kill all competitors, and remove all choice...earning MS a fortune that will make their current trasure chest seem like a pittance. Do we want to pay for that?

When 'free software' beats the final criticism into the ground -- that it is difficult to install and expensive to administer and maintain -- MS could start to lose miserably.

Developers have got to start makeing their user schemas intuitive and logical, and that takes user testing...I think we are seeing a lot more of this, and it is on the increase.


Treatment, not tyranny. End the drug war and free our American POWs.

Re:Not bloody likely

Ports depends on someone figuring out how to compile various packages on new systems - no centralized point of authority (or, more importantly, blame). Furthermore, there's no equivalent of a device-independent language for writing new applications.

Are the moderators on crack today or is this your troll account that you then mod up from your normal account when you have mod points?

Don't be ridiculous. The centralized point of authority for the BSD ports system is the BSD ports team. In exactly the same way as the Debian developers (and bug-tracking system) are your first port of call for problems with Debian-packaged software. Of course you can't really sue them if something terrible happens, but you can't sue Microsoft either - check out your EULA.

As for there being no device-independent language for use with the ports system, what do you think Java is? Or, for that matter - Perl and the Bourne Shell, which are almost universal throughout the Unix world? Sure, Perl is interpreted (although it doesn't have to be) but even as an interepreted language it shifts very quickly. Don't forget things like the GTK+/Perl bindings too, so don't argue you can't write user-friendly GUI apps with it - you can.

What would be an interesting project though is a JIT compiler for Perl - it has everything going for it otherwise as another, Open Source alternative to C#, including huge ease of use advantages.

Re:Why MS and .NET will win

Well some people are really into the .NET cloning idea, but for starters I don't think .NET will actually work for the desktop market, because of one very simple reason: bandwidth. The internet is simply not ready for service based applications. And certainly not for dbase intense applications that store the contents of your letters or documents.

.Net is all shiny and beautifull technology, and the concept of downloading the equivalent of dlls instead of jars must sound like Stravinski to Gates, but the hard cold truth is that it will atleast take 4 to 5 years before the world is done with struggling over bandwidth (if at all).

There's also another reason why I wouldn't recommend cloning .NET on linux. Many people allready suggested that keeping a clone up to date with microsoft's original is firstly a sign of weekness in the creative department, but more importantly it is sensitive to changes in the core, so that a clone will never be trustworthy (and thus market degraded). Businesses will never ever want a linux platform for running .NET. Thirdly, if MS should on the contrary decide to let linux clones thrive and florish, who will benefit from that ? The company that sells the services, because they have a cheapo .NET client and everything works as expected. Right. Now remember the Kodak story from a few posts back and you'll know that MS is not going to let this happen.

In short, I think if MS manages to get control over 3rd party services in some way, linux should not design .NET because it will be like a poisoned apple, a virus, allowing only more desktops to run native MS code (that gets things done, no hard feelings). If MS doesn't manage that level of control, then obviously they will break the clones by changing the platform specs, rendering linux desktops unusable in the business world.

Quite a dilemma. Ofcours the OpenSource movement could try to write it's own alternative platform now, in fact, it should be doable with such fine examples like c# and java, not to mention other languages which may be even more runtime-optimizeable and memorymanageable. On the other hand it is also worth noting that SUN has developed Jiro, Jxta and ofcourse Jini and JavaSpaces, which all focus on making J2EE a reality as a service solution. So in fact .NET, which is not funcitonal right now, allready exists, only under a few different API names. Maybe it's not a stuned as you would like, but hey, don't expect .NET being perfectly tuned and ready for another 2 years minimum.

Anyway, I've never really understood why the linux community would rather bash MS and run w2K alongside instead of opening their eyes and see SUN really doing a tremendous effort to take away a bit of the MS heat. Sure, Java may not be as sexy as your python, perl or c, but it the Grand Scheme of Things(tm), it is the best alternative anyone can imagine. They have bug submission, they have structures set-up to work swith large userbases, and they do deliver, allthough not OpenSource. I don't see anybody developing what SUN has done AND succeeding to outrun MS with OpenSource initiatives just yet. It's a painfull truth, but we should not kid ourselves and make sensible choices nonetheless.

Cheers,
Ignace

Re:What Ken Thompson thinks of Linux

By the way, BSD didn't exist in a Free form until after Linux was already started.

Yes and no. It did exist in "free form", then it got attacked by AT&T (for trying to be "free") and was in limbo until the courts and coders pulled the last bits of AT&T code out and created 4.4BSD Lite. Had this useless exchange never happened, it's very unlikely that Linux would have existed at all. -At least according to interviews I've read of Linus, stating how he'd probably have just used FreeBSD for study way back when...if it hadn't been encombered.

On other notes...
BSD didn't invent TCP/IP, but it did give Unix it's first incarnation through the Berkeley Sockets interface. First VM in a Unix system was also BSD. Reliable signals, BSD again. Fast File System, BSD. Passing descriptors over UDP, BSD.

Not that SysV just stood there. Most all ideas from BSD have been adopted by SysV, as well as many the other way (shared memory, streams, etc).

A HUGE part of what all Unix is today is directly traced to BSD work. Work from Linux so far, hasn't shown really up anywhere else. Don't get me wrong, Linux is a great system...but innovative it isn't. Thankfully, one doesn't need to innovate to be useful. -Hell, if that was the case Unix itself would be long since dead. :-)

Re:Simple breakdown

Dosen't apt-get install source xxxxx suck in the source, compile and install?

Yumm... Ports!

I love the FreeBSD ports system. To me, it alone is reason enough to run FreeBSD (though I run Linux). I'm glad to see it coming over to Linux systems!

If you haven't tried the ports system, I highly recommend you do so. They're a VERY convenient way to install new software.