Weekly Edition
Return to the Front pageSponsored link
Serve your customers, not your servers, with VERIO Linux VPS. Full-access test-drive here.
| Weekly edition | Kernel | Security | Distributions | Search |
| Archives | Calendar | Subscribe | Write for LWN | LWN.net FAQ |
Movable type and "almost free" software
That user base is rather less pleased now. With the announcement of Movable Type 3.0 came the news that, for all but the smallest, personal sites, use of the new version would require a paid license. Six Apart Ltd., the company which owns Movable Type, has since learned what happens when you upset thousands of people, each of whom has a personal printing press. Many online commenters have expended countless electrons on criticism of Six Apart and its new license.
We'll not join them. Six Apart owns its code, and sets the terms for its use. The company is behaving no worse than any other proprietary software vendor, and better than many. One might argue it should have made its new licensing plans clear before inviting beta testers to help them finish 3.0, but that's about it.
What Six Apart has done is to provide an object lesson in the perils of "almost free" software. If you do not have the right to run, modify, and redistribute a program, you will, eventually, find yourself in a situation where that program loses its value to you. If its owner fails to maintain it, nobody else can. If its owner imposes an onerous license, your only options are to take it or leave it. Source-available proprietary software can be deceptive; it feels much like free software. But every such package is another MT 3.0 waiting to happen.
Consider, for example, the case of qmail. It is, beyond doubt, a powerful and secure mail transfer agent. It is distributed in source form. But it also comes with a non-free license which forbids distribution of modified versions, and which makes the distribution of binary packages difficult. There has not been a new qmail release since June, 1998. Patches are required to get it to build on a modern Linux distribution, and others are needed to bring it up to the level of functionality needed by many sites. But, due to the redistribution restrictions, nobody can take over qmail maintenance and release a new version.
That notwithstanding, many sites (LWN included, it should be said) have chosen to run qmail. But all such users should bear in mind that qmail's license terms are, at best, vague; the software itself comes with no explicit license. If qmail's author were ever to proclaim a new license, it would be hard for users to prove that any other terms had ever been in force. Even without that sort of problem, it seems pretty clear that qmail's author has long since lost interest in working on the code; the chances of there ever being another qmail release appear small.
The Movable Type episode has shown, once again, that licenses really do matter. A free software license represents a sort of gift from a developer to users: those users will never be deprived of the right to use, modify, and distribute the covered software. Developers are not (and should not be) required to offer such a gift. But if the author of software you use has not given you those rights, you should not be surprised when the terms change in the future.
(Log in to post comments)
Qmail: There are great alternatives
Posted May 27, 2004 8:18 UTC (Thu) by stuart (subscriber, #623) [Link]
Well Exim works damn well and is GPL. The power (and more) of sendmail but with a configuration file. :-)For people who can't cope with exim / don't need its advanced features I am sure postfix will do although postfix's license seems a little iffy too.
Stu.
Qmail: There are great alternatives
Posted May 27, 2004 8:58 UTC (Thu) by alspnost (subscriber, #2763) [Link]
Well Stu, you beat me to it. I would also recommend Exim, and nope, it's not because I'm down the road from where it's developed ;-)Thus far, I've been able to avoid ever having to configure Sendmail. Hopefully, I can continue this course, thanks to Exim primarily. I'm actually delaying taking my LPI level 2 exams in the hope that the syllabus will be updated to cover alternative MTAs. Please, don't make me do Sendmail!
Anyway, I didn't actually realise that qmail was in that rather perilous state - I haven't looked at it for ages. So, when the time comes to change, it's got to be Exim....
Sendmail isn't that bad, also mt users: check out wordpress
Posted May 27, 2004 11:16 UTC (Thu) by fredrik (subscriber, #232) [Link]
Alspnost,
First, let me say that Sendmail really isn't all that bad as the rumor makes it out to be. The thing to remember is to use m4 to generate the configuration files, instead of trying to hand carve obscure instructions directly. Really, I started out with linux using sendmail and it wasn't until I moved to debian that I also moved to Exim.
Secondly, for all movable type users out there, let me recommend wordpress as a way to move to open source. It's rather automagic to install, and I've heard rumors of a script that can import mt posts to wordpress.
/ Fredrik
Sendmail isn't that bad, also mt users: check out wordpress
Posted Jun 3, 2004 10:44 UTC (Thu) by ingvar (guest, #1530) [Link]
Erm, surely introducing the m4 mess only made sendmail harder to configure? I could (while jacked up on enough caffeine and well-immersed-enough in line noise) write and make sense of sendmail.cf but no end of assorted tricks have managed to make me happy I understand of the m4 files actually work. Back when I had to maintain a sendmail.cf using the m4 files, I did the sensible thing and frobbed around in the m4, checking the config by reading the raw .cf file and frobbing again, if needed.
Sendmail cf files
Posted May 28, 2004 2:12 UTC (Fri) by LinuxLobbyist (subscriber, #6541) [Link]
Thus far, I've been able to avoid ever having to configure Sendmail. Hopefully, I can continue this course, ... Please, don't make me do Sendmail!
Awe, come on, now! I do ALL of my programming in sendmail cf rules. It's the creme de la creme of programming languages.
:-)
Sorry. I just had to.
I actually am a dedicated, diehard, happy sendmail user. As an MTA, it has certainly taken it's lumps in the area of security (though *much* better as of late), and in difficulty of configuration (though m4 has helped a lot). But it's a rare moment when I think of something I would like my MTA to do and I find that sendmail can't do it.
But...to each his own.
Support Freedon, only use Free Software
Posted May 27, 2004 12:01 UTC (Thu) by ber (subscriber, #2142) [Link]
So when will LWN switch to a Free Software alternative (like postfix or exim)and thus helping Free Software?
Support Freedon, only use Free Software
Posted May 27, 2004 12:07 UTC (Thu) by corbet (editor, #1) [Link]
"So when will LWN switch to a Free Software alternative (like postfix or exim) and thus helping Free Software?"
You know, our current configuration works; it also makes use of some qmail features that will complicate a transition. If I were starting now, I would install something else - but this is a mail system which has been running since 1997. I don't believe that our running qmail hurts free software, or that our switching would help free software in any notable way. But it would hurt LWN in the short term.
We'll switch at some point, it's definitely on the list of things to do. But I expect there are still quite a few things on the list which will get done first.
Support Freedon, only use Free Software
Posted May 27, 2004 13:43 UTC (Thu) by utoddl (subscriber, #1232) [Link]
We'll switch at some point, it's definitely on the list of things to do. But I expect there are still quite a few things on the list which will get done first.
...and might get done even faster once you unleash the hords of willing coders (at least half a hord anyway) by opening up the LWN.net code base as I've said over and over and over before.
Sorry; just couldn't resist. :-)
Oh, about the comment "The thing to remember is to use m4 to generate the configuration files, instead of trying to hand carve obscure instructions directly." Ah, right, like using m4 to hand carve obscure instructions indirectly is much better. I don't think so. Any config file where spaces vs. tabs makes a difference is broken, period. One so arcane that it requires an m4 front end to have any hope of getting it right is off-scale low. My $.02.
Releasing the code
Posted May 27, 2004 13:55 UTC (Thu) by corbet (editor, #1) [Link]
Releasing the LWN site code is certainly one of the things on the list that lies well above replacing the mail system. I've not forgotten it. But that release will require a significant block of time that I have not, yet, been able to find. Maybe this summer, if all goes well...
Releasing the code
Posted May 27, 2004 16:34 UTC (Thu) by lacostej (subscriber, #2760) [Link]
I personally would accept to not have news for a week if it was to get the source code afterwards...(Note I am a paying subscriber)
Make a poll and see how paying subscriber would accept to have such a deal.
Maybe extend the subscription one week to everybody?
Releasing the code
Posted May 27, 2004 17:07 UTC (Thu) by utoddl (subscriber, #1232) [Link]
Make a poll and see how paying subscriber would accept to have such a deal.
But, that would require implementing polls in the LWN code base, which would take time, but that could be alleviated by opening the source, but that would take time... Ack! I'm paralyzed by progress!
Releasing the code
Posted May 28, 2004 12:37 UTC (Fri) by lacostej (subscriber, #2760) [Link]
But, that would require implementing polls in the LWN code base,You can use an external service to implement the poll. That would break the loop.
Releasing the code
Posted May 27, 2004 18:46 UTC (Thu) by Klavs (subscriber, #10563) [Link]
I must say that I'm sorry you choose not to release the source of something that is GPL (AFAIK - and so is the changes you have done) - because it's not "pretty" enough (I believe that was your words on why you didn't just open it up).I (also a paying reader) would very much like it if you could just offer CVS access (through sourceforge - no hassle for you then) - so we could have a look at the code - and if you wrote feature-wishlist - perhaps some of us would like the features implemented and decided to do so. Using CVS it would not be too much work to clean up your code yourself (or us the readers just might) - and then merge the new features too.
Hope you see the light - with the code in CVS - it can't hurt - and nobody should be so vain - that there code changes to a GPL project, can't stand the light of day IMHO ;)
Releasing the code
Posted May 27, 2004 18:56 UTC (Thu) by corbet (editor, #1) [Link]
It's not a matter of "pretty enough." The LWN site code certainly has its grungy spots, but I feel reasonably good about it as a whole.The issues are things like doing a serious security audit (which, of course, I should do anyway), a minimum of documentation (the code has a lot of dependencies and is a bit tricky to bootstrap), and a determination of what parts of the code, if any, are to be withheld. I'm a little nervous about releasing the financial code, for example...I can probably get over that, but I have to look at it hard first. There is also the matter of supporting the code release: mailing lists, CVS repositories, dealing with patches, etc. Takes time.
Even then, of course, what the community will have is code to make a site that looks like LWN. One should not expect a general-purpose news portal construction system.
Anyway, it really is a priority to get the code out; it would be fun if others could use it and improve it, and I see it as simply the right thing to do. The problem is that there are a lot of priorities...
Releasing the code/priorities
Posted May 27, 2004 19:32 UTC (Thu) by utoddl (subscriber, #1232) [Link]
As a sort of self appointed needler about opening the LWN code base (at least I think I've dredged it up more than any other subscriber), I feel I really must step in at this point and state up front that I totally agree with you. Releasing any body of code has an enormous up-front cost in time and energy. I'm incredibly impressed with the amount of work you manage to get done -- the content, coverage, keeping the site up, etc. It's a fair question, whether the benefits you would receive from the occasional code snippet from us (who are so lazy after all that we've resorted to paying for web content already) would ever justify the effort and time it would take you to open the code as a proper project. My guess is that it might. Eventually. But it probably wouldn't feel like it for a while.
Anyway, I hope my occasional needling about opening your code base is taken in the spirit intended. Specifically, I appreciate your efforts and welcome the oppotunity to support them, whether financially in a small way, by word of mouth, or someday perhaps with some code. OTOH, lwn.net works as is, so perhaps contributed code is not really necessary for lwn's well-being right now. Continued production of quality content is, so that's a reasonable priority. Keep up the good work.
Releasing the code
Posted Jun 1, 2004 1:14 UTC (Tue) by macfisherman (guest, #6018) [Link]
I'd expect to pay more for LWN if the code is released. I think Corbet has the right priorities. I'm paying for the content, not the code.
Support Freedon, only use Free Software
Posted May 27, 2004 23:05 UTC (Thu) by nix (subscriber, #2304) [Link]
Oh, about the comment "The thing to remember is to use m4 to generate the configuration files, instead of trying to hand carve obscure instructions directly." Ah, right, like using m4 to hand carve obscure instructions indirectly is much better.Well, that's an excellent argument against, um, all compilers. After all, they're just tools that `hand carve obscure instructions indirectly'.
Now it's true that they often also go from a portable to a nonportable representation: but consider javac...
Another Grumpy Editor's topic?
Posted May 27, 2004 16:05 UTC (Thu) by southey (subscriber, #9466) [Link]
As with previous ones, this could be an interesting topic whenever you decide to do it. Your hands-on accounts are really useful and informative.
Another Grumpy Editor's topic?
Posted May 27, 2004 16:07 UTC (Thu) by corbet (editor, #1) [Link]
Trust me, I rarely miss an opportunity to recycle my work for multiple purposes...:)
Debian and "almost free" software
Posted May 27, 2004 14:27 UTC (Thu) by hazelsct (subscriber, #3659) [Link]
So where are all the Debian-bashers now? You know, the ones who complain about how "Debian is too fastidious about licenses and rejecting almost-free software!"Bring it on y'all, I know you're out there...
qmail
Posted May 27, 2004 15:32 UTC (Thu) by kurtm (guest, #4070) [Link]
Just a couple comments. You linked to the qmail.org site and not qmail's official site at http://cr.yp.to/qmail.html.Also, I've been looking at building an email infrastructure, and I've decided to go with qmail. When I looked at the release number and date, I thought immediately of abandonware. However, one trap I feel the free software community has gotten into is thinking that a lack of releases equates to abandoned.
Yes, a lack of security updates definitely makes one think 'abandoned'. But it seems that too often 'lack of more features' is thought of in the same way. Why do our tools need to keep getting more complex? Is a Craftsman Hammer useless if it's design hasn't changed in 20 years?
I *do* feel that djb should address the linux compilation issues. However, the rest of qmail *works*. The add-ons that lots of people use are just that, add-ons.
qmail
Posted May 27, 2004 16:29 UTC (Thu) by iabervon (subscriber, #722) [Link]
I think there is some value in the occasional insignificant tweak just to let people know you're still paying attention. With software, there's no effort to duplication, so you need to do something to let people know if it's still supported. With physical objects, they won't be for sale new after 20 years if they're not still in production, so you can tell that they're still available by looking in a catalogue. (And this could be worthwhile; Craftsman's lifetime warranty is less useful if they stopped making new ones 20 years ago). I think it's not so much new features that people want as signs of life. (Unless, of course, there are particular new features that they want)
qmail
Posted May 27, 2004 17:41 UTC (Thu) by kurtm (guest, #4070) [Link]
So djb should essentially just bump the version number occasionally so people know the project is supported? I don't understand otherwise what 'some value in the occasional insignificant tweak' would mean.Unfortunately, I think 'signs of life' in this case means 'convenient signs of life'. There is a rather active qmail mailing list. People don't want to do more work than looking at the version number (and maybe when it was released).
Yes, often no releases in a while can be an indicator of abandonware, but it is just an indicator.
I think there are two traps that get us to think this way: 1) Most proprietary software companies (and hardware companies for that matter) have trained us to think that stuff more than 18 months old is useless. and 2) Because so much free software is being released before it is ready and/or while it is still evolving that we assume anything that doesn't make new releases all the time is not being maintained.
qmail
Posted May 27, 2004 18:17 UTC (Thu) by corbet (editor, #1) [Link]
On the other hand, when there have been no releases for six years, and the current release does not even compile on a modern Linux system, many observers must wonder if the code is being maintained at all...
qmail license
Posted Jun 4, 2004 19:52 UTC (Fri) by bodosom (subscriber, #3774) [Link]
There's been endless wrangling in the past about the qmail license and what it means. Although DJB hasn't made much new code recently he has updated the documents to clarify certain things.1) You can't distribute a modified qmail without his permission. Most distros don't want to deal with the anti-LSB DJB so they don't seek permission.
2) You can distribute a package (e.g. deb, rpm, pkg) that changes qmail in the post-install. This requires a build enviroment (compiler etc.). Some folks consider this an unwarranted imposition.
3) You can distribute binaries that give the net result of building a set of pristine sources. You can also make useful links to those binaries.
4) New Linux (the errno problem) is unsupported. This is awkward but not fatal given 2.
The key thing is that while you can create a derivative work you cannot redstribute it (at least not and call it qmail). His license is not like other free/open source licenses but you can achieve the same end result.
I'm just saying... (rather tardily too).
qmail
Posted May 28, 2004 14:17 UTC (Fri) by mmarsh (subscriber, #17029) [Link]
"Convenient" signs of life can be important for someone deciding whether to install some program. It doesn't have to be feature creep or arbitrary version bumping. Some sort of notice on the project's homepage would suffice: "[28-05-04] We're still looking at including nifty feature X", or "[01-01-04] This product is now considered stable. We have no immediate plans to add new features, but bug reports are always welcome and we'll keep an eye out for needed security fixes. If there's any change to this status, we'll document it here."I was recently looking at alternatives to CVS, and one that caught my eye was OpenCM, which has what appears to be a unique cryptographic authentication feature (unless you want to run Apache behind Subversion). The latest release, however, is an alpha for 0.1.2, and that was almost a year and a half ago. It took a fair amount of digging to find out that this _isn't_ abandonware, it's just going through a major restructuring. That makes me feel a lot better about considering it, but it took awhile to figure out. That's time that could have been spent elsewhere to greater benefit.
qmail vs free software
Posted May 27, 2004 17:01 UTC (Thu) by cdmiller (subscriber, #2813) [Link]
We have used qmail for about 6 years now. It's rock solid, secure, and easy to configure. We have 16 lines of configuration in the 7 files of our qmail/control directory on the main mail server. My default exim.conf for a debian/GNU workstation has 200 lines of configuration. The postfix setup on one of our redhat servers has 76 active config lines in main.cf.It's interesting to watch something like postfix or exim evolve alongside a fairly small and static product like qmail. The free software products in this case appear to become more complex and feature ladden over time.
My organization can't redistribute modified versions of qmail, but we don't need to. We can patch our version to our hearts content if need be, and patches are merrily produced and distributed by the community.
It would be interesting to see what the community would do with a free version of qmail. Maybe DJ Bernstein will be convinced to allow a free software fork of qmail under a different name one of these years. Even then, qmail will be hard to beat for it's ongoing record of stability, security, and ease of configuration.
Postfix
Posted May 28, 2004 1:28 UTC (Fri) by brouhaha (subscriber, #1698) [Link]
I use qmail on a server machine for about a dozen domains. It works reasonably well, but I'm disappointed in the lack of ongoing development. There are patches to implement many of the new features I want, but the commonly requested features really ought to be part of the distribution, and it appears that this is not likely to happen.For all my backup MX systems, I've switched to Postfix. Postfix has a feature that makes it incredibly easy to set up a backup MX, which is that you can configure it to be willing to act as a backup for any domains which include specified IP addresses in their MX lists.
In other words, if I have a machine at 192.168.34.56 that acts as a primary MX for a bunch of domains, I can simply tell postfix on the backup MX system to act as a backup for any domain that has 192.168.34.56 in its MX list.
I also now use Postfix on a few leaf systems, and it seems to work quite well there.
So I'll probably move my main server from qmail to Postfix at some point. The transition will be somewhat painful, as I currently depend on some qmail-specific behavior in .qmail files that won't easily map to Postfix .forward files. Fortunately Postfix does have a configuration option that can be used to change the subaddress delimiter from it's default value "+" to another character such as the "-" used by Qmail.
Postfix
Posted Jun 3, 2004 0:03 UTC (Thu) by bronson (subscriber, #4806) [Link]
Hey, thanks for mentioning that!*bronson adds 192.168.34.56 to his DNS tables as a backup MX.
F/OSS is not a gift . . .
Posted May 30, 2004 20:54 UTC (Sun) by Kope (guest, #21434) [Link]
A free software license represents a sort of gift from a developer to users: those users will never be deprived of the right to use, modify, and distribute the covered software.
I have to disagree wtih this statement. An open license is not a gift, it's a warranty -- and the only warranty that really counts. It warrants that the user will never ever be deprived of their right to use the software they have choosen to invest in for whatever reason, includng possible future capriciousness on the part of the developer.
To say it's a "gift" mistakes healthy cooperation and partnership with gratuity. A gift implies that the developer is doing something that is not required of them by propriety. That is not the case. A software user is entitled to the freedom to use the software into which they invest their time and money. Often customers take signficant risks in adopting software packages, the wrong choice can be ruinous to an enterprise. It is only in guaranteeing that the customer will never be deprived of their own data and the functionality to access it does software gain value. Proprietary licenses do nothing more than retain to the developer (or copyright holder) the freedom to abuse the customer. Not retaining that right is not offering a gift, it is restraining from abuse.
Further, to articulate that such licenses are 'gifts' falls into the rhetorical trap of accepting that the way to make money from software is to sell restricted licenses. It is to adopt the worldview of the Microsofts of the world over the IBM's and Redhat's. Something advocates of Linux should never do intentionally.
Entitlement?
Posted May 31, 2004 20:28 UTC (Mon) by shane (subscriber, #3335) [Link]
A gift implies that the developer is doing something that is not required of them by propriety.So you're saying that when I spend my personal time developing software I am required to release it as open source?!?! Do you have any idea how insane that sounds?
How about this as an idea instead: If I spend my time writing software, then I get to decide what happens to it.
I fought with my employer so that I could contribute to open source projects on my own time (as opposed to the typical "intellectual property" agreement which puts grants everything I do to them), and all of a sudden here you are dictating to me how I can license it!
If you, as a user, decide that you will only use open source software, that's your right. But that's all.
I'm a strong believer in open source in general, and Linux in particular, but your attitude upsets me greatly. Open source software is often a labour of love, and should be treated as the precious gift that it is.
Not fair! qmail, Mozilla Firefox ...
Posted Jun 3, 2004 13:34 UTC (Thu) by job (subscriber, #670) [Link]
The qmail bashing is completely unfounded. qmail is not less free thanMozilla Firefox!
You are not allowed to redistribute modified versions of any of them.
Firefox protects this with their trademark license, qmail in another way,
but it is the same restriction. This is because the authors feel afraid
that modified versions would not be up to their standards. In my personal
opinion, the fear is unfounded, but we have what we have and these
packages are good, almost-free, software.
There are indeed new versions of qmail from time to time, but it is
called net-qmail. It is not hard to find, just go to qmail.org and look
under "download". You will find that the software when compiling applies
patches to an (included) standard qmail source tree, much like LAME did
in the beginning, but that is a technicality only.
These licenses where you are allowed to develop but not do formal
releases, that may or may not conflict with the author's, are not
uncommon. I mentioned Firefox but LaTeX has a similar license. It may be
a bit sad, but bashing one in particular won't help. You need to stick to
facts in order to have a meaningful discussion.
Not fair! qmail, Mozilla Firefox ...
Posted Jun 4, 2004 18:21 UTC (Fri) by bayard (guest, #20385) [Link]
> You need to stick to facts in order to have a meaningful discussion.Isn't it true that the only trademark encumbered parts of Firefox are
Netscape brand items (such as the "N" logo) and Firefox can be built
without them, losing hardly any functionality?
That is the impression I get reading the 'copyright' file in Debian
mozilla-firefox package, anyway. Would you care to clarify?
Not fair! qmail, Mozilla Firefox ...
Posted Jan 3, 2006 7:51 UTC (Tue) by rickmoen (subscriber, #6943) [Link]
"job" wrote:qmail is not less free than Mozilla Firefox! You are not allowed to redistribute modified versions of any of them. Firefox protects this with their trademark license, qmail in another way, but it is the same restriction.
You can fork Firefox, i.e., distribute derivative works in either source or binary form and use them for any purpose, provided you don't call it Mozilla or Firefox. That's what Flock is, after all.
qmail's generous but proprietary licence does not permit that. For example, the netqmail authors have (commendably) done absolutely the maximum possible to make the problem minimal, but neither they nor anyone else (except Prof. Bernstein) may lawfully distribute it in any binary form.
It would be nice if qmail advocates would content themselves with having a well-performing and generally secure if aging MTA, and not continually attempt to stretch the truth past the breaking point.
LaTeX has a similar license.
Not exactly. Knuth said: "The program for TeX is in the public domain, and readers may freely incorporate the algorithms of this book into their own programs. However, the use of the name TeX is restricted to software systems that agree exactly with the program presented here." (His typefaces have a similar restriction, and some country-specific hyphenation-pattern files are restricted. It's a rather muddled situation.)
The LaTeX code itself, aside from encumbered pieces from Knuth and some other TUG members, is open source under LPPL v. 1.3.
Rick Moen
rick@linuxmafia.com