Slashdot Mirror

The MySQL model was not radically different from how Aladdin Ghostscript used to be released. Stallman tolerated that licensing model as a necessary evil. So it would be non-coherent for him to be against the MySQL model. RMS is a pretty cohesive guy in his ideas. He is also more flexible about revenue creation models for free software than people give him credit for.

I'd be really curious to hear what Peter Deutch (Aladdin Ghostscript) and the commercial SSH developers have to say about idealism, commercialism, earning a living, competing against their own earlier free software, etc.

Not sure about the SSH developers, but there is a good interview with L. Peter Deutsch at http://devlinux.org/deutsch-interview.html.

Ghostscript comes to mind; there are two separate licenses: the AFPL license requires a fee and is necessary if you are commerically distributing Ghostscript. If not distributing commerically, you can instead use the standard GPL license. The author has done very well for himself using this "dual license" model.

More pointedly, I would direct you to an Interview with L. Peter Deutsch which addresses the precise issues surrounding copyright assignment that you seem to think so daunting.

Ghostscript has been not finding them to be a problem for a lot of years now.

The question seemed to me to be if it was possible to protect the code from being manipulated. If they have well-defined, open interfaces between everything with a high level of modularity, it might be possible for some person (but probably a company) to use many of these modules with some replaced by their proprietary modules. The product as a whole would be proprietary, though many of the pieces would not be. This would obviously be in violation of the spirit of the GPL.

The GPL does try to protect against this very thing. A GPL program cannot link -- even dynamically -- to anything less free than the GPL -- or, conversely, linking to a GPLed library can only be done by something that is GPLed. There is a (possibly dangerous) exclusion for system libraries -- but in a Win32 environment, that exclusion is obviously quite necessary as well.

It's not clear whether the GPL's protection would also would be true for other interfaces -- what is a shared library? Is a CORBA interface a form of shared library? Is it protected under the GPL? Other interfaces? I think this is the issue in question. The output of a GPLed program is specifically not covered under the GPL, but output and interface can be a fuzzy distinction at times.

Some people question whether you can protect an interface. That is, does fair use protect the use of all public interfaces? I believe the author of Ghostscript had these very same questions. You can read an interesting interview with him where he talks about these things and why he used a different license.

One thing you might do is to clarify your own interpretation of the GPL in this context. This is something Reiser did for ReiserFS -- and he got a lot of flack for it, because he was adding something the the GPL. But I think it makes sense, since Linus has made his (I think flawed) interpretation of the GPL in regard to binary-only modules, you shouldn't let that be seen as the de facto interpretation in regards to your software.

The part of the reiserfs faq that talks about it not being able to be remounted without an fsck only applies to the current, non-journaled version. The whole idea of a journaled FS is that you can mount it instantaneously (compared with ext2) after a crash. The faq you were looking at hasn't been updated to reflect this yet.

FAQ we're referring to: http://devlinux.org/namesys/faq.html

It also has it's own version of fsck. I copied the following strait from thier sight

"Can reiserfs be used as a root partition?

Reiserfs can not be mounted if it was not cleanly umounted and it had been mounted read-write. It cannot even be mounted read-only. You have to boot some other way to repair the reiserfs partition first using reiserfs.fsck. These ways can include: a read-only reiserfs partition used just for the booting phase, RAMFS, and an ext2 root partition. Please share your experiences with doing these so that we can add to our installation guide."

Taken from

http://devlinux.org/namesys/faq.html

Reiserfs is blazing fast for rm -rf. About ten times faster. than anything else available. Don't believe me? Check these benchmarks yourself.

ReiserFS looks like it has a similar problem. It can't be mounted *at all*, not even read-only, if it was not cleanly unmounted (see the FAQ). So it can't be used for a root partition either.

Ext3 will probably be the replacement for ext2, since they are fairly compatible. Distros will probably bundle other filesystems and make them available for non-root partitions.

Though benchmarks aren't everything, they're always nice to look at.

Here's the linkage:
http://devlinux.org/namesys/bens.html

--

Hard disks are about 3 cents per megabyte.

RAM is about $1 per megabyte And according to rumour, IBM is about to speed up the capacity race in hard disks, while memory is about to hit the quantum effects wall.

RAM isn't a real alternative to disks and if you want to use RAM to cache disks, then you need to be very careful that your access patterns are such as to let the cache have an effect.

Research should be into how we make a large amount of RAM and a humongous amount of disk space work as fast as if the disk were RAM. Perhaos ReiserFS will help Linux with this.

Incidentally, DAT tapes aren't much cheaper than hard disk space these days, though they are more portable, which is important (if you back up by copying to different disks within the same LAN then you have a problem if lightning strikes your ethernet).

Not to mention Hans Reiser's fs called (guess what) Reiserfs.
It wouldn't have journaling at first, but it is planned.

Isn't this sort of the idea behind devLinux?

LinuxWorld published my analysis of the SCSL and I mention a few things about the SCSL that I like on my dev/Linux site.

Stig
--
dev/Linux (ORGanized, NETworked, COMmunity)