Meeks: Measuring the true success of OpenOffice.org [LWN.net]

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 10, 2008 15:49 UTC (Fri) by forthy (guest, #1525)
Parent article: Meeks: Measuring the true success of OpenOffice.org

OO.o is not "sexy" for developers. What itch does it scratch what a "real programmer" would need it for and therefore improve it? I write my documents (papers, presentations) in LaTeX, using LyX as frontend if appropriate. I draw my graphics still in xfig (inkscape would be an option, if the developers would care more about technical drawings, like better grid, component library, better arrows, and especially a user interface which doesn't hide this deep in the attic, etc.), I render my diagrams with gnuplot. And I can quickly write programs that massage data much better than a spreadsheet. I tried oocalc for that purpose, but it's sluggish when drawing graphics, and it's such a repetitive task to make the calculations - I won't do that again (this is a fundamental design flaw in how spreadsheets work).

Furthermore, a lot about OO.o is to make it work with Microsoft's so-called "document formats". This affects design decisions within how OO.o works - and especially affected them in the past. Reading and writing a format full of cruft and legacy already makes a program bloated and full of legacy, not a fun to work with.

IMHO when you dump a load of formerly proprietary code onto the free software community, it won't take off. Or at least it requires that everything is rewritten once or twice before people accept it (like Firefox).

to post comments

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 10, 2008 19:12 UTC (Fri) by tzafrir (subscriber, #11501) [Link] (7 responses)

The Linux kernel isn't sexy either. You need to reboot the system to activate changes, for starters. It's easy to completely hang your system. It has its own build system and its own set of libraries. The stack space is limited to 8k or even 4k.

Most of the contributors to the Linux kernel are corporates, not individuals. I suspect they do it not because the Linux kernel is sexy, but because fixing something or adding a new feature into the kernel would eventually help their business.

Now aren't there companies out there that need OOo improved?

Specifically, IBM is always mentioned as a big contributor to OOo.

As for "Scratching the itch": what's the rate of the presentations in recent Linux-related developers conferences that were made in OOo?

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 10, 2008 21:50 UTC (Fri) by neilbrown (subscriber, #359) [Link] (2 responses)

Most of the contributors to the Linux kernel are corporates, not individuals.

Are you sure? Checkout "Greg Kroah Hartman on the Linux Kernel" on video.google.com. About 24 minutes in there is a nice table of where the contributions really come from. Number 1 is "Amateurs"

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 10, 2008 23:01 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]

Yes, more precisely it was hobbyists or people whose organizations affiliations were unknown or they didn't want it disclosed followed by Red Hat, IBM, Novell and others. If you add those up, you can easily see that he large majority of contributions are from commercial organizations.

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 11, 2008 5:11 UTC (Sat) by tzafrir (subscriber, #11501) [Link]

No. 1 "contributor company" is "(None)". But those unaffiliated individuals are still responsible for only 15%-20% of the code. What about the rest?

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 12, 2008 20:15 UTC (Sun) by dvdeug (subscriber, #10998) [Link]

How does that make the kernel not "sexy"? In terms of the original poster, the kernel is sexy because it's necessary and something that every developer uses. In my terms, it's sexy because there's nothing quite like playing that close to the hardware. All those limitations add to it being sexy; there's no room to drop a megabyte array on the stack, you can't just toss in libbloat to simplify a problem, if you screw up, it's going to crash your system and possibly trash your filesystems to boot. Difficult and dangerous is very much sexy.

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 13, 2008 8:30 UTC (Mon) by forthy (guest, #1525) [Link] (2 responses)

what's the rate of the presentations in recent Linux-related developers conferences that were made in OOo?

Well, those are made by the people who didn't master LaTeX beamer ;-). Last time I saw a presentation from a RedHat employee (a gray-bearded man with a pony tail), it was made with PowerPoint. People who do this sort of things are happy even with the limitations their tool provides. If you stop being happy with ooimpress or PowerPoint, you first look around if there's something better, something that makes your presentations look more professional (and that certainly doesn't mean "more silly animated effects"), and quicker to create.

My conclusion is that when you want to write a new "office suite" from scratch as free software, first make a good extensible typesetting and drawing engine ([La]TeX has some limitations, so be better than that - especially the programming language it uses is a nightmare; PostScript as drawing engine is not perfect, either). The user interfaces for the different programs should be frontends to this engine; they need to be extensible, as well (packages for the typesetting engine need to tell the GUI something, too). Make sure that writing an extension is easy, and that the foundation is stable and sane so that most of the work goes to writing extensions. Concentrate on good practice, don't try to imitate the bad user interface of the competition.

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 16, 2008 14:15 UTC (Thu) by kamil (guest, #3802) [Link] (1 responses)

No offense meant, but I think you must be very young...

I'm with you on the use of LaTeX for document writing. I too use gnuplot for my graphs, but I did switch from xfig to inkscape (the GUI of the former would drive me nuts).

I made a few presentations in LaTeX, but then I switched to OOo. Why? Because of issues with sharing. Slides are shared a lot more than documents. In a collaborative environment (aren't they all?), people exchange slides from each others' presentations all the time. Unless you don't have any co-workers, or they are all as huge LaTeX fans as you are (in my experience, unlikely, even among programmers), using something radically different for your slides than the rest of the crew is a big PITA. OOo is close enough to PowerPoint to make importing and exporting practical, and that really helps.

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 25, 2008 16:27 UTC (Sat) by linuxrocks123 (subscriber, #34648) [Link]

I also use LaTeX for presentations, and my approach to the sharing slides problem is to first convert other people's slides to PDF (if they aren't already in PDF), then create a Makefile that uses GhostScript to splice and reweave the output of SLiTeX with the slides I want to import.

If someone wants to use my slides in PowerPoint ... well, they have the PDF, and PowerPoint can import PDFs as images at least, right?

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 10, 2008 22:18 UTC (Fri) by Wol (subscriber, #4433) [Link]

Furthermore, a lot about OO.o is to make it work with Microsoft's so-called "document formats". This affects design decisions within how OO.o works - and especially affected them in the past. Reading and writing a format full of cruft and legacy already makes a program bloated and full of legacy, not a fun to work with.

And that explains only too clearly why I dislike OOo (and PAY to have a copy of WordPerfect on my system - that said, I don't particularly like the new versions of WP - ever since the v9 rewrite!). I find Word a counter-intuitive mess, and I find OOo too similar to Word for comfort (and WP is moving ever closer to Word :-(

Cheers,
Wol

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 11, 2008 0:31 UTC (Sat) by jasonjgw (subscriber, #52080) [Link]

The only circumstance in which I would consider using OpenOffice.org is that
requiring collaborative editing of a document with a user of Microsoft Office.
I am glad OO.O exists for those people who need such interoperability, and I
would use it should collaboration with MS-Word users become necessary.

However, having experienced the benefits of LaTeX for a decade now, I don't
want to go back to using a word processor, which would be a big step in the
wrong direction. I also suspect that the barrier for entry into LaTeX macro
package development, for instance, is lower than that of OpenOffice
development.

Important lessons ought to be learned here. The next major typesetting and
document processing system needs to be designed with extensibility and
flexibility in mind, so that it can attract extensions and improvements just
as TeX, Emacs and Mozilla do. A graphical interface should be optional, as it
is in the TeX environment. Interoperability with HTML and XML will be more
important in the future than legacy, binary formats issuing from Redmond,
Washington.

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 11, 2008 5:15 UTC (Sat) by tzafrir (subscriber, #11501) [Link]

If one wants to scratch an itch with OpenOffice, one may find it simpler to write an extension:

http://extensions.services.openoffice.org/

So I suspect the statistics would not be complete without looking at the extensions.

Is there any effort to co-maintain most of them together with OOo? Compare that to external kernel modules and to Firefox plugins.

Meeks: Measuring the true success of OpenOffice.org

Posted Oct 11, 2008 8:04 UTC (Sat) by robilad (guest, #27163) [Link]

<quote>Or at least it requires that everything is rewritten once or twice before people accept it (like Firefox).</quote>

Spot on.

I'd suspect that the reason for the Linux kernel code churn lies in part in encouraging rewrites of existing functionality to make sure new generations of maintainers & developers have code sections they feel attached to very quickly. And it works. ;)