OSNews

Hold the phone!!!!

It looks like there were build failures. People will have to wait a few days before this is ready, I suppose.

Sorry about that.

For Arch Linux OpenOffice.org 3.1 was available two or three days ago:-)

I'm not exactly sure what is going on here, but I did get OpenOffice installed in Kubuntu after a couple of tries.

http://matt.bottrell.com.au/archives/350-OpenOffice-3.1-for-Ubuntu....

I had to use apt-get from the command line. Everything worked as expected except the last step "apt-get upgrade".

When done from the command line, that step reported that openoffice.org-java-common was not installed. So I did "apt-get install openoffice.org-java-common" and that worked. After that, the "apt-get upgrade" step worked.

When I first started OpenOffice under Kubuntu, I got the horrible Raleigh appearance theme. Yuk.

Now I had previously solved the problem of appearance of GTK applications in KDE4 (Kubuntu) by installing gtk-chtheme.

http://plasmasturm.org/code/gtk-chtheme/

I had used gtk-chtheme to select the Qt4 theme for GTK applications, rather than the default Raleigh. Worked like a charm for firefox and other GTK applications.

But OpenOffice was now ignoring it.

So I started up KPackageKit, and searched for OpenOffice. I saw a package that had this description:


So I manually installed that as well, and then it was all sweet. OpenOffice now correctly displayed itself in the GTK+ theme that I had selected with gtk-chtheme.

Searching for the sarcasm tags ...

[s] Nah! Warp drives are clearly more important! [/s]

Well, the official explanation is that items on page 2 don't have a "More" section, not that they are less important.

Well, needless to say, I think that it sucks and page 2 is an awful idea. I liked it better when all items were on page 1 and a few were contracted.

I completely disagree with their reasoning as to why things are on the first or second page.

For them it is about the amount of content, not what the content is about.

So ... if someone figured out the meaning of life and it took less than 50 words, despite the possibility that everyone might suddenly be happier than ever thought possible, it would go on page 2 just because there wasn't enough text in the article.

For example, the perfect answer is 42. We don't know what the question is but with an answer being 42 there isn't enough text to make it onto page one. So anyone that might know the question would be unlikely to look at page 2 and therefore wouldn't be able to let all of us know the question.

Yes, it is assbackwards reasoning. The front page concept has a time-earned meaning to people and just arbitrarily redefining it doesn't work. Front page means "important" news, not minor uninteresting stuff like, say, the remote possibility of warp drives.

I've noticed that whenever "worthy" (what some people may call "worthy") items appear on the front page, hardly anyone gives a hoot. It's protocol. Nothing special.

...but when "worthy" items appear on Page 2, people come out of the woodwork just to criticize OSNews, its editors , Thom (there are more editors aside from Thom, and often he gets the brunt of "mistakes" that other editors have made, so I suggest at least checking the name of the author of the article in question before going on a rant towards a specific editor if a rant is what you're after), the president or prime minister or whatever of their respective countries, the Easter Bunny, and their fathers. It sometimes seems to me that some people thrive off of criticism and spreading ill-will.

I don't know about you, but the release of OOo 3.1 isn't something I'd bother about whether on the fist page or the second page. Yippee. Release. I'll get it when I'm ready. It's really all a matter of perspective, and sometimes there are issues that may be important to you but aren't quite as important to others... happens all the time in life. I find that we just have to deal with it. If you think an issue is important yet others don't, you have the choice to either let the issue pass, whine about how others don't think the same way you do, or take action and get the issue more in the open. Which is most effective? Take your pick.

**sigh** I really oughtn't to even have to write this in the comments on any article.

Edited 2009-05-08 22:15 UTC

The question is what is 2 x 21

At least that's what my printer started spitting out one day in hexadecimal, so that's my story and I'm sticking to it.

Now what does that mean? Either the drinking age in the US (and the "life" implications of what might happen after a night of said drinking), or nothing at all!

It is certainly very siginificant news. But only news that one of the editors knows enough about or cares enough about to write a "Read More" article gets on the front page. Brain dead, I know. But since the editors have officially appropriated OSNews as their own collective personal blog, there is not much we readers can do except keep an eye out for another similar site whose editors have a better attitude. (Suggestions are most welcome.) Any criticisms are summarily deleted. So you read this one fast.

To be fair it was 2.4ish that they got rid of that horrible slow startup. Interestingly they have got rid of the speadsheet slowness, not something I have experienced myself, but those who seem to know how to find faults point it out.

For a *minor* version it seems like a perfectly reasonable thing to mention a *minor* feature.

I suspect from the work done it would have been better to follow firefox's path of naming it 3.5

Feature pages are simply a list of features someone has chosen of note. I quick look at Gnomes latest release of Vistas Release you see similar items like "paint has 10 undo's instead of 3"

The reality is that little niggles in any interface are annoying, the fact that the OpenOffice.org team are addressing these is a good thing

Edited 2009-05-08 15:00 UTC

On first start after boot OpenOffice writer takes 8 seconds to load fully. On subsequent starts, it takes about 2.5 seconds to load.

New features, half a million new lines of code, yet it starts faster. Uncanny.

"

Under Ubuntu 8.04 LTS on my even more modest system, the lshw command reports the following:



On first start after boot OpenOffice writer takes 8 seconds to load fully. On subsequent starts, it takes about 2.5 seconds to load pretty much the same as with your system.

Half as many processors a quarter the memory yet it starts as fast. Uncanny.

Word 2007 on my quad-core Windows 7 machine: 1 second.

Word 2003 on my Aspire One with windows 7: less than a second.

Still, OpenOffice's improvements sound impressive. Too bad it's bogged down by an interface I simply loathe. I prefer 2007 all the way.

Windows Vista, and I presume Windows 7, aggressively pre-loads memory with stuff it might need.

http://www.codinghorror.com/blog/archives/000688.html



http://en.wikipedia.org/wiki/Superfetch#SuperFetch

This of course means that Windows 7 and Windows Vista will be slower to boot, but much faster to load any applications that Superfetch decides to pre-load libraries for.

You can take it as a given that libraries that Office uses would be high on the priority list for superfetch.

So how to compare apples with apples?

If people wish to achieve short load times for OpenOffice on Linux, and they do not mind longer boot times, then this is the go:

http://www.techthrob.com/2009/03/02/drastically-speed-up-your-linux...



When Oo writer was taking 15 seconds to load, with the preload running daemon running it was reduced to about 7. I'd imagine then that the preload daemon could cut OpenOffice 3.1 start time down to about 3 seconds (on my modest machine). Possibly a bit less with some specific tuning.

Best of all, the preload daemon in preference pre-loads programs you use most often, not necessarily programs written by the same people as those who wrote the OS.

http://en.wikipedia.org/wiki/Preload_(software)

So, does anyone have some timings for starting OpenOffice 3.1 on a Linux system with the preload daemon running, once they have trained preload to believe that OpenOffice 3.1 is a popular application?

Here on OSNews, after all, we like to make perfromance comparisons on at least a kind-of level playing field, don't we?

Edited 2009-05-10 13:27 UTC

You don't understand this at all.

The goal of a desktop machine is to deliver me the applications *I* use the fastest. Windows is smart enough to keep track of the apps I use, and pre-load those that I use the most. In other words, it anticipates my needs, and acts upon it. THIS is what a computer is supposed to do: make my life easier. If I see too much RAM not in use, the software is not handling it right.

Linux is apparently unable to do this for me, which means longer loading times for applications, which means longer interruptions in my workflow. That is not what I want out of a desktop machine.

There IS a level playing field. It's just that Windows is smart enough to prepare that field for MY use, and Linux does not. What you are saying is that it if team A beat team B, you'd be saying the competition was unfair because team A trained more, and prepared better!

That would be ridiculous, right?

No, you don't understand this at all.



Linux IS able to do for exactly what Superfetch does for you. All that you need to do is to install a single program called preload. This program is not typically installed by default, but it is readily available in most distribution's repositories.

apt-get install preload

and it is done.



There is not a level playing field comparing Windows 7 (with Superfetch) starting MS Office to Kubuntu or Ubuntu without preload installed starting OpenOffice.

That is what you effectively did with your post:


Because they are on Windows 7, those start time utilise Superfetch. The start times quoted earlier for OpenOffice were without similar assistance from the Linux equivalent functionality, called preload.

Once you install preload on a Linux system, which then has similar functionality to a Windows system with Superfetch, only then do you have a more level playing field for such comparisons.

My goodness Thom you fly off the handle quickly ... and so often incorrectly too. You really should try to control that temper of yours a bit more.

Edited 2009-05-10 13:57 UTC

And again, you're being nonsensical.

Does any major Linux distribution come with preload? NO.

Does Windows 7 come with prefetch? YES.

Since you need to compare performance figures using DEFAULT, up-to-date installations, it is PERFECTLY normal EVERYWHERE in the world to compare app start times the way I did.

Obviously, it's still anecdotal, but that's besides the point. You are claiming that pre-fetch gives Windows an advantage. Well DUH! That's the whole point! The point of an operating system is to use the hardware it runs on to its FULLEST potential, to make MY life easier.

Apparently, as even you admit here, Windows 7 does that better out-of-the-box than (K)Ubuntu does. And that's the point. We'll see what happens when Linux distributions start shipping something as advanced as prefetch by default, but for now, Windows is the BETTER operating system in this regard.

And even you yourself admit it.

Sigh!

Both Superfetch and preload offer users a COMPROMISE. They compromise a slower boot time to give users a faster load time of often-used applications. Neither one helps at all for seldom-used applications, but they still take up the time at boot and make for a slower boot time.

AFAIK, Superfetch is built in to Windows 7 and Vista, so Windows users aren't given a choice. They must endure the longer boot times to have the faster load (only for oft-used applications though).

The default on Linux is not to install preload. That is the opposite default choice to Windows. However, on Linux, users have a choice. If they want preload, they can just install it. It will operate automatically for them from then on. If they subsequently decide they don't want it, they can remove it. If they change their minds, they can re-install it again.



Yes, it is. It is normal practice in media to ignore the longer boot time of Windows, and to claim a faster load time of applications. It is also normal practice to ignore the fact that users can make exactly the same compromise on their Linux systems if they choose to, but instead to claim that Linux necessarily has a longer load time for applications and ignore the faster boot time.

This does not mean that the normal practice is in any way sensible or fair. It just means that the pushers of such FUD wish people to spend their money on Windows.



I'm not claiming that Superfetch gives Windows an advantage ... I'm claiming that Superfetch compromises a slower boot speed on Windows for a faster time for application loading of commonly-used apps. If you prefer it that way and you want to make that exact same compromise on Linux ... then you should install preload.



It doesn't do better. Windows takes up the time required to get data off disk at boot, instead of at application load. As I keep trying to tell you, if you prefer that behaviour ... then it is trivial to get Linux to behave in the same way.

But just straight out ignoring a slower boot time and claiming a faster application load time, because that is the compromise that Windows happens to offer, different to the default on Linux, is typical of Windows deception (not telling the whole story) when it comes to performance comparisons.



Sigh! Preload is quite equivalent to Superfetch.

Windows is only better if you insist on ignoring half of the story.

Anyone with any sanity at all would recognise that the better system is the one which gives end users the choice of where they wish to make the compromise that these utilities both employ.

Edited 2009-05-10 14:45 UTC

Under Ubuntu 8.04 LTS on my even more modest system, the lshw command reports the following:

...

On first start after boot OpenOffice writer takes 8 seconds to load fully. On subsequent starts, it takes about 2.5 seconds to load pretty much the same as with your system.

Half as many processors a quarter the memory yet it starts as fast. Uncanny. "

It probably depends quite a lot on disk I/O, but still ...

However ... my system was running Kubuntu, and yours was running GNOME. Since OpenOffice is more Gnome-ish than Qt-ish, I would presume my system had to load a lot more libraries than yours.

I'd really be interested if someone would look into a Qt front-end version of OpenOffice and Firefox. That would be good, IMO.