Saturday December 10, 2005; part of: OOo
Below the fold are extracts from some emails passed on by the Guardian.
UPDATE: The last word on open source usability, from Dan Kegel's notes of a conference on this subject:
Q. These usability videos are great, thanks!
A. They're the only chance some Linux developers have to see women. (audience laughter)
Arnaud Fortier (no relation) writes:
What you say is so true !! I'm on linux because I can't stand cracking / using software which I don't own the licence, this is how millions of computers runs windows.
But I'm a bit angry about egocentrism and stupidity in Opensource software Who can fix bug in kernel ?!
And what about the fact that there are more and more distributions of linux each one incompatible with another What is the point of changing names, places of config files or more cut them into pieces ... And why are software installed everywhere but in the default place !! I don't know
why this appends, but it's agrat opportunity for other Os, and in the same times it results in stupid stuffs like :
- different package for each distro !!
- no way to find a correct answer for your own distro unless you are on
great ones like Fedora / Mandriva / Suse ...I can't imagine the amount of time spending by all those so-called geeks
to recompile, repackage for each distros !! What a waste of time
Thanks anyway for that kind of waking-up article ( but don't expect too much wake up ... )
Sorry for my bad english I'm a poor french guy ;-)
Sebastian Rahtz writes:
Firstly, what differentiates open source is not the development model, but the licence under which we
use it, not the development methodology. Many of the concepts described by Raymond's "Cathedral and Bazaar" metaphor apply to the world of Agile software development, used in commercial contexts as well as open source.
Secondly, the distinction between open source as "amateur programmers", and closed source as "professional programmers" is seriously misleading. Open source development is largely done by computer scientists, either students or paid practitioners. The difference is that they are contributing to a common core of material with a robust licence instead of competing with each other.
Many of the contributions which do come from outside Sun are from exactly the kinds of users Andrew describes as "the non-programming public" and dismisses. These are the localisations of OpenOffice into Welsh, Scots Gaelic and a multitude of other languages, which are one of OpenOffice's key strengths.
Darryl LeCount writes:
I found Andrew Brown's vitriolic attack on OpenOffice.org to be ill-informed, heavily biased against open source software and extremely inconsistent. He claims to "like" OpenOffice, initially using it out of "a mixture of perversity, stinginess, and vague anti-Microsoft sentiment", before launching into a tirade about how buggy it is and how flawed the open source model is. The author has clearly neither had extensive experience of using Mozilla Firefox, Blender, or Linux, and it is also clear that he has had little involvement with the development of
these products despite his vague claims.
[acb comments] I use firefox from time to time. I just don't like it as much as Opera. I have experimented with four different distributions of linux, and even ran FreeBSD for a while. It's good server software, but none who was not a hobbyist, a masochist or a crank would use it as a desktop. As for the development of OpenOffice, I have, let's see, written the macro which was for two years the only way to get a count of selected words. I started the UK spelling dicitonary localisation. I have written macros to do title case, and to work around the badly broken move-by-sentence function in OOo 2. I have submitted s great many bug reports, and checked other people's. None of these are large contributions. None of them affect the core code. But thousands of people have found them useful to judge from my download figures. Anyway, back to Darryl ....
He is correct in stating that millions of users use OpenOffice.org, and that only a small proportion of them can do anything remotely useful with the source code. However, let us suppose that of the 50 million downloads he quoted, 30 million are still using it. It is safe to say that at least 1% will be able to do anything constructive with the
project - that is already 300,000 users and any of them could - and often do - have something to contribute, be it a bug report, bug fix, translation, graphic work or some other addition to the program. The Joint Copyright Assignment agreement only applies to code submissions.
The marketing claim was of 50 million users and far more downloads. Why is it safe to say that 300,000 people have contributed? Where are these 300,000 contributors? Where are their contributions? So far as I understand it, the JCA (which I have signed) applies to anything on the web site, and anything distributed with the programme.
Alastair Stevens writes:
I have just read Andrew Brown's musings on OpenOffice* in this week's technology supplement, and I'm compelled to disagree with his conclusions.
I'm been an OpenOffice user for some years myself, and I agree that it has its major flaws, and that its development pace is more glacial than many would like. However, it is well known that the open source model doesn't always work well for certain classes of software, this being one of them. That conclusion is nothing new.
But to generalise it into a sweeping slur on the open source development model is completely wrong. Open source has more than proved itself in the arena of infrastructure software; after all, vast portions of the Internet's servers have run on it for years.
I've never argued that some Open source programs don't work well. Python, Ruby. PHP, MySQL ... My point is that the method only works when developing stuff that programmers, rather than users will use.
Eric Raymond (yes, that one -- there is a quote from a libertarian politician in his signature) writes:
In fact, controlled comparisons between closed- and open-source versions of functionally equivalent programs have been done. Barton Miller's well-known "Fuzz Papers" suggested that open source programs to have a 39% edge in reliability over closed-source equivalents. So where are the comparative statistics for the bug load of Microsoft Office? Do we know that it has fewer than 11,000 bugs and feature requests outstanding? If Mr. Brown don't know that, or at least have those figures for a closed-source program of comparable size to OpenOffice, he has no basis for asserting that the open-source method is failing.
Peter Ross writes:
I read Andrew Brown's article and I think he has a number of faulty premises. One is that an assumption of the open source world view is that "a significant proportion of users can fix bugs". Clearly only an extremely tiny number can fix or wish to fix bugs. But when they do the results can be made swiftly available to all, and if Andrew Brown needs some bug fixed in a hurry he can even pay an independent expert to do that.
There are bug bounty schemes. They don't work. A decent programmer -- even hired for one day -- is going to cost me a lot more than a copy of MS Office.
Jan van Leewen writes:
Do you not read german articles or reviews?
OOo has been put to the test next to MS-Office and some other suits.
It was concluded that both can do about all the jobs you want to do with an office-suite.
No one was more buggy then the other except for the corel-wordperfect-suite that had some minor bugs.
Now I have not read about tests from you so what is your opinion based upon?
As one of the supporting forum-members said: it is not the question if OOo can do this or that but how.
Look -- there are, right now, more than 700 bugs officially listed in 2.0 which are supposedly fixed for 2.01. None of mine are in there, so far as I know, because none of mine are due to be fixed for ages. Here's a nice simple one for you. type a paragraph of text with several sentences. go to the end. Press shift-ctrl-backspace to delete one sentence backwards. What happens?
Posted by andrewb at December 10, 2005 01:06 PM
Thank you! As far as I'm concerned, OpenOffice.org, even at 2.0, is a piece of nearly unusable junk! I work for a company that runs a Windows-based network infrastructure, even though our main product is a Linux-based server. :)
I'm a mathematician and programmer, and I run almost exclusively Linux at home. I've had occasion to attempt editing corporate Word documents on my machines using OpenOffice. It flat out doesn't work half the time. It messes up the fonts, usually to the point of no return. And, as the reviewer at ZD pointed out, it's slower than Microsoft Office.
In addition, the "PowerPoint" clone can't do a decent job of rendering PowerPoint presentations. I gave up trying to create presentations with it. I don't have much experience with the OpenOffice spreadsheet, though. As a programmer and mathematician, I use more "serious" tools like R and Maxima for most of the things regular folks do with spreadsheets.
The sad thing is that for document generation and editing on Linux, there are so many open-source tools that are much better than either flavor of Office! Unix was born a text processing environment for scientists and engineers, after all.
I can do so much more with LyX or TeXmacs, I can write PostScript and PDFs with dozens of tools, make much better diagrams with XFig than Visio, do a better job of desktop publishing with Scribus than Acrobat, etc. About the only thing I still need Windows for is MindManager and TurboTax.
There are many hundreds of quality open source applications out there. I'll take the R statistics package over its multi-thousand-dollar commercial competitors in a heartbeat. I'll take the KDE desktop over Windows. I'll take the LAMP stack over Windows, IIS and SQL Server. And I'll take LyX and TeXmacs over Office.
Yes, there are many hundreds of quality open source applications -- OpenOffice.org isn't one of them. It's an embarrassment to the concept of open source. OpenOffice.org just sucks.
There are some useful remarks in this article. I can well believe OOo is more suitable than MS Word for writing books, but that's faint praise indeed. Even after all these years, the Word table feature still corrupts content and formatting at random. The incurable weakness of OOo is that its purpose is to be a drop-in replacement for Word, and that forces it to reproduce a lot of bad concepts, flawed design, and outright bugs. Personally, though, I care a lot less about non-working operator interface shortcuts than its stability in working with native ODF documents. I have no need to deal with legacy documents in MS formats.
My choice for large writing projects is FrameMaker; unfortunately, it's too expensive to equip a whole organization with it, and its native file format is nonstandard and undocumented. Scribus is its intended successor, and I hope it succeeds in growing into that role. Last time I checked, its table feature was still in too early a state of development for serious professional work, and almost everything I do involves multi-page tables.
Can I just say a word about OO from the perspective of British schools. We have long had a problem. We cannot afford to invest in more than one commercial Office system (and then only because we get a heavy discount), nor can we train pupils and staff in multiple systems. Most govt and training documents arrive in Word, Access or Excel formats. OO being free (also free to us as Star Office) means that we can add it to our systems as an alternative. This is especially true now that Base has been added. The look and feel of the applications can be adjusted to without a major change of orientation and most of our documents are simple enough to allow at least some transfer.
Pupils and staff can also, at last, be encouraged to put an Office system on their home computers without breaking either (a) the bank or (b)copyright laws. The timeliness of this, as broadband communication makes file transfer between home and school simple, could really contribute to a breakthrough in the cross-curricular use of ICT in our schools. However, whether OO actually replaces MS Office on our systems depends on the resolution of the problems you have been discussing.
I have read the comments in the Guardian regarding OpenOffice 2.0 and disagree. The comments not only unfairly disparage OpenOffice but open source software generally. Given the fact that open source does not have the marketing dollars of a Microsoft, negative comments about open source do a disproportionate level of damage, that's why open source developers react so angrily.
Any bugs in OpenOffice are counter balanced by the fact that it is free! That might not mean much to some people but to someone in the developing world, free software is a significant public good. Furthermore, at $600 per workstation, is Office really all that much better?
As for problems formatting a .doc document, this is an argument in favour of developing non-proprietary document standards, not for sending more of your hard earned money to Bill Gates.
I'd like to see an article in the Guardian about how Microsoft's security vulnerabilities go unpatched for months at a time. In the meantime, try reading this article from the Harvard Business Review comparing Open Source to the successful development model at Toyota. Access to the full article requires a fee. Hmmmm why must knowledge have a price?
Mr. Brown's observations are on target, but the analysis isn't. The problem with open office, is similar to the problem of X or mozilla. The packages were formerly developed by a very small group, cathedral style, and only became 'open source' as a strategic decision very, very late in the game. At that point, decisions which made sense for a small group having to maintain something become obviously bad ideas for a project to have large appeal:
- the project uses it's own build system.
- @ ooo1 no native widgets were used on any platform.
- the project is monumental and self-referential.
- the licensing was wrong-headed (JCA.)
These are all bad characteristics for open source code, because they make the project much bigger than it needs to be, and uses no common parts with any other projects. That was a good thing for proprietary code (fewer dependencies, less licensing, easier maintenance.) but people familiar with other open source projects face an extremely steep learning curve, and are thus discouraged from participating.
The OOO people understand this, and you can see that by the directions taken: In OOO2, the developers moved to use of native toolkits. The use of the inhouse/licensed adabas database used in v1 is replaced with a layer to provide access to other projects' work such as mysql. The email client and desktop in window ui's of the old Star Office were also jettisonned. What you are seeing is code in transition. hopeful next steps are:
- break up file format support into libraries, so that it can be shared with other office projects. something like libopendoc, libmsoffice, libmsxml, etc...formats become pluggable,and separable projects from oo.o itself.
- The build system needs to become more 'normal'. These are activities that will make the project more approachable for open source developers.
It is a similar process to what has happenned in the past year or so with transition from Xfree86 to x.org. While XFree86 was open source, it was run Cathedral-like, with patches being accepted from only a chosen few. After a lot of frustration, essentially a revolt took place, and the X Server (the equivalent of a Display driver and a low level set of widgets on windows.) moved to a bazaar model. The software is being broken up into small components, the unique build system is being regularized.
Open source craves small replaceable components with well-defined interfaces, so that implementations can compete with each other and replace eachother. There are no small components or well defined interfaces to openoffice today. Releasing code is not a pixie dust which magically transforms the architecture of the code to and permit instantaneous vast improvements in development speed.
Besides who you share the code with, Open Source also means:
- build only what you have to, leverage the best technologies out there for components.
- When you need to, make small, independently useful components.
- put them together elegantly to make an integrated package.
The open office team, from the actions they have already taken, clearly 'get it.' but it will still take them years to get there. They have to transition a huge proprietary monolith into something modular, featureful, and sexy for developers.
Beyond that problem is a much smaller but not sinsignificant decision to use copyright assignment. Contributors to the linux kernel have all kept their copyrights, and someone who would like to take the kernel private would have to negotiate with several thousand organizations. Sun wanted to maintain control over the licensing, which is legitimate on their part, but it clearly lowers the value of the code to potential contributors. Once a single organization is set up as copyright manager, only that organization need be swayed to make changes.
So In the end, to use openoffice as an example of what open source doesn't do well is a bit unfair because a number of it's problems stem from its proprietary history. OpenOffice is still in the midst of overcoming that legacy. By 3.1, as you hypothesized, it should be vastly improved.
[Matthew Begbie's original comment lightly edited for presentation]
MB writes: "Any bugs in OpenOffice are counter balanced by the fact that it is free!" Can I interest you in this chocolate teapot? It is free! No, it doesn't work with boiling water, but why put money into pockets of kettle manufacturers? Surely the answer is for everyone to make tea with cold water.
OOo is more use than a chocolate teapot. But this isn't because some Microsoft software also has bugs. I really wish OOo worked better. I can see the advantages, to schools, to governments, and even to individuals, if it did. But it doesn't.
I have been involved in many 'cathedral style' product developments, including one that has been described as 'the most successful commercial software product in the world'. The interesting thing we discovered was that having a very large number of bugs on the list was a measure of product quality, not the opposite.
This is because bad products don't get used (no bugs), mediocre products get a small number of bugs reported ('I just need this to get it to do this one task') but great products get used for lots of things all the time. Most of these are not scenarios the developers considered testing and so generate bugs. So I don't think it is necessarily a bad thing to have a large number of outstanding bugs.
However, I have tried to use OpenOffice to replace MS Office (I have both on my computer). I like OO and use it daily. But it does feel a little flaky. Even though it is clearly getting better, it hasn't replaced MS Office on my computer yet. I think Andrew's article clearly expresses why he has the same problem.
Rather than attack what he says (which I find needlessly defensive) why doesn't the Open Source community take his remarks on board and suggest ways to make OO better?
Peter Silva is the only comment writer I can relate to here - he tells us that OO is going to drop its own middleware layers to use those available from the environment. This is a very good idea - as would be KDE and Gnome adopting Mozilla Firefox as their browsers.
I think the author of the article has achieved exactly what he intended to do and that is generate traffic to his blog and article. If you were a good objective writer you would not need to resort to this tactic. It’s a bit pathetic that you feel the need to be so negative at the expense of something you get for free. Let's face it, this article could just as easily have been positive but that just would not have generated the traffic right :-(
And now I feel annoyed that I even responded :-(
You can forget the Guardian as a valid source of information on Open Source. It and its technology blog do nothing but constantly trash Google, Apple, Linux and anything else non Micros~1.
For someone who has contributed macros he sure sounds like he don't like anything about Open Source. Does he have anything to say on his own personal use of OpenOffice. What exactly did it not do to his own satisfaction.
Did Brown do any real research before writing this diatribe. Of course he's going to produce the contents of his inbox as evidence of Open Source 'kindergarten noise. Talk about getting your retaliation in first. That whole article reads like nothing more than troll bait. Lets see some choice quotes.
"OpenOffice .. vividly demonstrates the limitations of open source as a way of producing software, and its futility as an ideology"
"I like OpenOffice. I used it long before it was usable, out of a mixture of perversity, stinginess, and vague anti-Microsoft sentiment"
"I have written numerous macros"
Which makes you eminently suitable to comment on the developers methodology.
"I have done quality assurance work, submitting reports on bugs and testing those reported by others. So I know something about the open source "community" and the enormous gap between myth and reality."
Tell us exactly what is wrong with the Open Source methodology. Who among the development community would support your views.
"The myth of open source rests on two improbable assumptions. The first is that a significant proportion of users can fix bugs"
A bit of a distortion. The claim is many eyes can spot bugs - not fix them. Why do you see the need to misrepresent such a basic Open Source position. Is the truth not sufficient to carry your argument.
'This is important because of the second crucial false assumption: that even if not all users can fix a bug, they can help find them. They can't. Most users just think: "The computer isn't doing what I want."
It is false because no-one ever said that all users can fix bugs. You're merely engaging in a classic strawman argument. Just as an exercise and assuming what you say is even true. How does a set of non bug reporting users invalidate a set of bug reporting users.
"Big commercial software companies .. have an incentive to avoid errors in the first place"
You mean like the LaSS.exe bug shutting down Windows and preventing the users from downloading the fix.
"Where's the support desk for OpenOffice?"
If you bought it from Novell for instance then you get the support from them. Otherwise its Usenet or your local users group. I'm surprised you don't know this.
"Despite the open source rhetoric, OpenOffice actually started as a commercial product - StarOffice"
How does this fact invalidate in any way Open Source methodologies. How many other projects actually went the other way and why do you deem it necessary not to mention them.
Incidentally, nice one invoking `rhetoric'. Tell us is there anything other than palaver coming out of OpenSource projects anywhere on the planet. Anything at all.
"just before version 2 was released, a Ziff-Davis blog . pointed out that OpenOffice is bigger, less efficient, and much slower than MS Office"
The alleged evidence being the worst case scenario the blog author could produce. A sixteen sheet 1628 rows by 13 columns 20MB spreadsheet. The author also failed to mention that the entire spreadsheet would not be available for calculation immediately in msOffice despite the main screen displaying some cells.
As for his claim that OpenOffice is a memory requirement hog. It is well known that msOffice loads much of its functionality at boot time therefore giving a false impression of load times and memory usage. As such the blog author would be well aware of this but chose to mention it. So much for the myth of msOffices load times.
Windows 2003 loaded modules:
Loaded modules at boot: 194, 45,544,901 194 MB
For Microsoft Office is loaded: 10, 26,513,732 MB
For OpenOffice is loaded: 77, 48,336,384 MB
"the shrieks of outrage in the comments would be alarming even in a kindergarten."
At which point we wave goodbye to your reputation as a serious commentator. One of the worst tactics of a Usenet troll a classic ad hominem if there ever was one.
"So why is OpenOffice so dire? The project claims more than 50m downloads of the software, so let's assume that 50m people have tried it at least once."
FudInjection alert: OpenOffice is dire. The project only claims as to the number of downloads. People download it but don't use it after trying it once. This Andrew Brown knows for a fact as people who do use Open Office haven't told him about it.
futility as an ideology
open source rhetoric
shrieks of outrage
"open source projects... rely on a very small group of programmers .. who have no direct incentive to work on the bugs that are important to users"
"many eyes making bugs invisible"
"I like OpenOffice."
translation: I really want Open Source to get better :)
http://technology.guardian.co.uk/weekly/story/0,16 376,1660763,00.html [guardian.co.uk]
ps: I posted this to the Guardian letters page but of course they didn't print it.
It is odd that Open Office is critised for not working the same as MSWord. How many MSWord users add new lines to get to a new page, or repeat a row in a table so that headers appear on each page? Most people I know that write small documents could do just as well by using Wordpad.
Experience from my childrens education is that they are NOT taught IT, but are taught how to use MS software - badly! When trying to help, I was dismissed for not formating the titles straight away, and for not correcting a spelling mistake as soon as I had finished a word. Presentation, not content, was the prime objective.
Instead of citing Open Office / MSWord, why not compare Firefox against MSIE? Firefox does the web browsing, and users can install 'Addins' for extra functionality. This is a good model for Open Source development. As with the origins of UNIX, it is design to be modular. A user selects the tools they need for the task at hand. A different user may select different tools for a different task.
The difficulty with any large project is keeping it understandable to its users. I buy a newspaper and throw away the fashion bit, my wife throws away the business section. When Open Source adopts this principle it can win easily, but when it copies a bad model it naturally fails.
Textile formatting works here. Double hyphens are automatically converted to en dashes, quotes are automatically smartened. You can put dashes and asterisks around text to make
and other silly effects easily.
- Text wrapped in Asterisks which * will be bold. The asterisks must touch each end of the bold text. There must a space before the first and after the last.
- Text wrapped in underscores - _ - will be italicised. The underscores must touch each end of the italics. There must a space before the first and after the last.
- Paragraphs starting bq. will be block quoted. There must be no space before the "b" and one space after the full stop.
- A hyperlink is made by wrapping the link text in double quotes, followed immediately by a colon, then the URL. If there is a question mark in the URL, wrap the whole lot in square brackets.
- I use two classes to mark up text that deserves it. sane text looks like this. loony text looks like that. The syntax for those is %(sane)[space] sane text %; loony is left as an exercise to the reader.