 Llywodraeth yr ydw i'w ddweud y fawr y byddai ar gyfer y llywaf. Ac gallwn gweld fawr yn tynnw i'r llywaf, ond yn ddiddordeb y tŷ. A'r llywaf yn tynnw i'r llywaf, ac yn ddiddordeb'r llywaf, ac â wnaethau'r llywaf yn tynnw i'r llywaf IWGI ar bod ddod oeeddi'r eich harfa unrhyw gŵr i chi. Mae wnaeth i'n ddiddordeb y bwysig? I ddweud eu bod yn ystod y troi deall, gallwn rhaglaeth fynd ar gyfer teimlaeth rhaid. Mae'r hwn yw yw. Dwi'n meddwl, rydyn nhw'n wneud y bydd wedi bryd. Rydyn nhw'n meddwl o'r mynd i'r brifau, dwi'n meddwl, mae'n ddim wedi'n meddwl wedi'i fyddio'r maen yn rhaid. Rydyn nhw'n meddwl yw, mae'r rhaid o'r 2.0 o ddysgu. Rydyn nhw'n meddwl, rydyn nhw'n meddwl'i dyfodol ar y pryd sydd ymgyrchol, yn y bwysig. Mae'r rhaid i'r bwysig yn ei gadw i ddweud yn y pryd sydd yma, Shall we see what what is the point? As they didn't know their name. Either way someone imaginative in the marketing department decided this new name, Base, for the database, So anyway, there's a bug in base. But all of this code was originally there but no one knew it was there and no-one was using it. So it's a real shame to invest such a huge amount of effort but you can only find it if you go file something something, sub-menu something else and then you're there. So we thought as a marketing rebranding and beautification. eraill y bydd chi fydd yn penharu i'r llch beth. Here it is. As you can see, this is a pretty screenshot of the evolution integration. It just so happens that this is done as a database back-end, a K-mail has a similar thing. You can do queries on your family addresses. It will show you my personal information. Which is great. Some other things that we improved in T-Zero are ergonomics. If you'd used impress to do presentations before, it was absolutely appalling, really, really unusably bad, and it's much less so now, and more to the point we have some very nice transition stuff happening, so you can do all these effects, you'll notice in my talk there are almost no effects at all, but some people when they get boards writing their presentation, they think that they can add value to their talk by adding funny transitions that then take a long time to happen while they're standing here and they can't flip sides quickly. But, you know, this is a bullet point that's important to have for marketing reasons, and now we have it, which is good. And you can select your master page and fiddle around with that. Spreadsheets, we now have 64k rows and some funky new pivot table things. I've been told to demo the pivot table functionality. This talk can become interactive. Right. Who knows what a pivot table is? I do, so I put my hand up, right? OK, good. Well, so pivot tables are very powerful, and almost no one understands them, as you can see from the straw poll, and so I'll just show you a very quick demo of that if I can find it, because it's actually useful. Let's have that one. And hopefully it'll load fairly quickly. So I have this highly confidential Navell sales data done from my database here, and as you can see we're doing pretty well on the eclectic things. But anyway, you see there's a whole load of data, and it's really unstructured, right? This is a very typical database output. And you can't really see who sold the most slugs, right? Or what happened in EMEA, or whatever. So what pivot tables allow you to do, which is really, actually I should point out at this point, since this man is recording it. But what I'm referring to when I say pivot table is actually not a pivot table, because pivot table is a trademark of Microsoft Corporation, which are very wonderful. What I'm actually showing you is something that could be like a pivot table, but it's actually a data pilot. I thought it was to make that clear and acknowledge the trademark. Good. So here you come to data or tools even, and hopefully you can find it. And where is it running? Tell me. I have this cheat sheet here. I think it must be in data. Data pilot. Yeah, here we go. And there we go. Start. And use the cancellation. Now what it allows you to do is to view, for example, the operative on this field here, so you can see, you know, which of these boldfellows are in March. And the region is column, and then maybe the number of units they make, right? So you can see very quickly who sold what in what region and how much money they got for it. And of course you see it's the sum here, so it sums all of those fields together. But of course you don't have to have a sum. You can have all manner of horrendous different kinds of things, you know, here's some standards and standard deviations and percentages of other things. And you know, it's incredibly powerful. And then of course you can filter it as well. So you can say we've already got a region. Let's say a product here. So when we create this guy, we get probably something meaningless at the bottom here. And you can see, then, the operative on this side, and you can see the region, and you can see the sum of everything. This is really basic spreadsheet itself. But since none of you understand it, I thought I'd explain it a bit. So that, you know, you can then see quite how useful this can be to visualise huge unstructured day sets. And of course it becomes more interesting when you see what the competition does later. But anyway, so this is the number of sales sold in these regions by those guys. Understand? Good. I'll shut up. Well, thank you. Thank you. I'm for my next trick. Good. So a whole load of stuff happens in writer and across the board to improve the ergonomics and to add a whole load of things. So one of the nice things that happens is that Microsoft adds a new feature into Office. And often it doesn't become very prevalent because it's not very usable. But then they add a whole load of usability stuff and people start using it everywhere. And so one of those things is auto shapes. And people start using them more and more and more to create graphs and drawings and diagrams and all these sort of things. So we add another feature and I'm treading on another trademark here. I believe auto shape is a trademark. I should probably label that too. So we have custom shapes. Okay. And so here are our custom shapes that are interoperable and compatible and blah. So you can draw more smiley faces more quickly and you can adjust them once you make them. Well, I can demo that at the end when I've run out of things to say. Floating dockable toolbars that pop up and you can move around. Format copy brush so you can copy styles from places and enrich your document more easily in a stylistic way. So instead of going bold, you copy a style from something like that. Nested tables obviously. Since the web was invented, people like to do almost every layout thing with deeply nested tables. And so it's nice if you can nest tables. And so, you know, this helps to interoperability a lot and people that like tables. And lots better desktop integration. Lots of game KDE, game VFS, KDE PIM, evolution file selectors and so on and so on. And here's a photograph of it. I say that's the next slide. That's the photograph of this clip art. If you if you have a penchant of drawing strange things, then you can add them to the clip art collection and we'll render them to bit maps and shove them in open office. And this is the, this is the latest beautiful most sexy search in the file selector thing, showing, you know, your nice game integration file selector and search for something. And there you can open it, which is good. So some other new stuff in open office to zero is X form support. I never knew what X forms was until I started using open office and saw it was a bullet point. X forms apparently is the next generation of revolutionary web content, right? It's good, isn't it? How many new X generations there are of revolutionary web content? I don't know. I mean, there are lots of them. Whether it ever comes to anything, I hope it does, I guess, but you can never tell. And but either way, there's a pretty gooey and you can draw the things nicely and it all does what it's supposed to and its standards compliance and all that good stuff. But at the end of the day, having filled your form out, you can write a little plug-in that then prints all of this data out. The US spends $15 billion a year mainly on student labour, I assume. So maybe some of you have a real feeling for this, typing that stuff back in again, you know? So you type it all in, print it out, post it and someone else types it back in again and then it's in your database, which is good. So the thought is that instead you can write a little plug-in that will print out a 2D barcode which you then sign and it turns out it was an indemnity clause, a mortgage on your house and you've just sold everything to someone else or something, which is good. Document signing. Okay, so also there's this wonderful world of cryptography which is really difficult to use. If anyone has ever tried to get a secure key that's cryptographically authenticated with other... Debian have these amazing things. It reminds me, you know, where you will sign keys and look at passports and all this wonderful infrastructure around signing that makes it almost impossible to use in a real-world situation, it's unusable. But it's all there, you know? If you've done all that stuff, it can work nicely. So I can't demo it. I'm not in on it. The Oasis file format, I'll talk a little bit about that later. It's a new standard. There are so many standards to choose from. It's brilliant, but it's a new standard. We have nicer PDF export, better interop across the board as a photograph here. The tiresome thing about interoperability is you can improve it a whole lot but there's not really much to show, you know? You can say it didn't work in the past and now it does, but hopefully no one noticed. So I'd like to tell you a little bit about OpenOffice itself. It's a project and why it's important. So there's a real price gradient here, okay? So normal people use it today. If you're hacking on OpenOffice on the train, as you do with your laptop running Linux or BSD or something free, people will go and show you, oh, that's OpenOffice. You're like, who are you? You go to a party with normal people in the evening and you wander around grunting at yourself and it turns out that loads of people are using OpenOffice randomly in their offices. This is brilliant. Of course, there's no way to measure it, but the people who are paid to wander around doing this sort of thing, they make up big numbers of millions, you know, eight million, or that's the number of lines of code. Hopefully we'll get a user for each line of code and then we can turn them into programmers and they can maintain that line of code. Anyway, but the nice thing is that if you look at the price of a desktop machine with Office on it, the pie chart breaks down a little bit like this, okay? So I rang Microsoft up to try and get some numbers out of them and they were not very helpful. They didn't believe I was an OEM, they wanted this partner agreement, they wouldn't give me any numbers. So I tried eBay and Google and this is the sort of thing I come up with. Most of the cost of the PC that you buy is Office. It's $12 billion for Microsoft and Revenue 2005 from the economists, 30% of their revenue. This is a really big product and we do a huge amount of what Office does or the things that people use from Office right now for free. Now, there are lots of other sexy projects out there that you can go and work on as Beagle, as Firefox and this sort of thing but if you look at the desktop stuff but if you look at the real value in this pie, none of these things really show up. Windows is quite cheap actually. Internet Explorer comes free as part of that pie chart. So Firefox is cool and it's a great piece of free software but in terms of the real wedge driving free software onto people's computers, OpenOffice is where it is you and I know that this is not the world's most polished beautiful piece of free software, right? So if we want people to have a good positive image of where OpenOffice is, I need some help. You knew that already, right? So where is freedom best served? If you write 100 lines of code today, if you write it in a tesh or something, you'll get a beautiful result, I'm sure you'll improve something there but it will be seen by a best, a fraction of the world's mathematical type setting community you will love it, right? The equation editor in OpenOffice, all of these poor students laboring all over the world will really appreciate what you did and there's a lot of easy fixes to do. Just as an example of raw money, I know money doesn't motivate everyone but Novell saved the million dollars in Microsoft Office licensing fees last year, we plan to save it this year and the next year and so on and so on. So it's quite substantial, everyone has it on their machine, 80% of them use it as their main office suite because someone persuaded them to say that in the survey. So this next slide is really the sort of wall that's hotting up between ECMA and OASIS or not actually between the standards groups because mostly the standards people don't mind, they're just about the technical solution but between other people that sort of hang around the edges and create walls, wonderful. So in case you didn't know, Microsoft are standardising their file format and switching to this new XML file format that they're doing for this published XML standard that anyone can participate in. If you want to club together, form a non-profit and join ECMA, you can do it. You can sit there and the amusing thing is if you sit on a technical committee, one member, one vote, so Microsoft has the same votes as Novell in this specification. Fascinating, isn't it? Anyway, so there's a whole lot of work going into making it and you can argue about whether it's beautiful or not. So this is more like the Microsoft non-mix content layout by the open document, how it does it, of course, very small so it fits on the slide, right? But you get the idea, the content is either you split up your whole document and make little spans that have various attributes or you build that into the actual flow of the document. But the interesting thing about the two is really that one is aiming at 100% backwards compatibility with Microsoft documents. That's the ECMA format and that is their overriding goal. They wheel out big banks who say, we quite like our spreadsheet to calculate the same value. Behind the scenes, a whole bank has run on spreadsheets and it's not very stable really and if anything changes, we'd never know and I didn't say any of that. But if you go to, interestingly, spreadsheet risk analysis conferences, banks' entire capital adequacy ratios which means how much cash they've got to have sitting around is being changed due to spreadsheet errors. Like this cell was wrong. Oh dear, we need another however many million dollars in cash. So this is kind of not cool if you come along and say, oh well we've got a new file format unfortunately none of your data comes into it and all the values are different. And some of these things are being addressed in open documents and whatever. We'll see what happens but customers don't want their documents to break. So that's my take and this is a controversial one that I've had by virtually anyone else I talked to so you can flame me after this. Son is handling open office star office. So another thing that people ask is what is son doing in terms of open office? How is son related to this? Aren't they screwing it up? Aren't they the big evil people? People are always looking for the evil person. They like to see the world like that. Actually son is doing a really good job of one thing which is licensing and getting this done in the open and they release all of their changes to the star office out there. They commit it into the public source tree and that's the authority base. They add a few modules. They have proprietary bits that they put on as well. But what you see there is essentially what is in son's product as I understand it. And they now offer support for open office of Chisbury. There's some legal bits. There's a joint copyright assignment so you own the copyright on your work but son does too. It copies your copyright. And the source is under the LGPL and the GPL obviously. And we've got rid of the sizzle because it was widely abused. You know, having this recording is a shame. In some ways, never mind. It was widely abused by some people. I would suggest you go and Google for fork open office and see who's been doing it. This is a very large company that's rather unpleasant about it. Anyway, another interesting IP issue is that son has got a patent covenant. A patent covenant is a really revolutionary. I've never heard of them before and they're cool. So, if you enter RMS's talk earlier, you'll hear what it's possible to say about patents if you don't have any money and you're not affiliated with any large company and so on. I recommend his wise words at the beginning of the talk to you. It's all good stuff. But what son's patent covenant essentially says is, thankfully, it says, this is the layman's view. I'm a layman. It says, we haven't done any patent research. We don't know if there are any patents that either we hold or anyone else holds. That's a good first start. That clears your ass. But if we happen to have any patents, which we don't know if we do, covering open office, we promise that we won't sue you if you use open office and you develop on it and you reuse it under the terms of licence. That's an incredibly powerful tool because it's very cheap to do. It's easy to write. It's non-controversial and it's brilliant. Microsoft has started to do this again for their standards and this, hopefully, is a brilliant tool in this IP hole clout. I hope that it will become more prevalent as a way to drive your standards and get it adopted. The approximate developer breakdown, as you can see, most of the people are at sun for full-time paid developers. I compensate and create a community segment there. Yeah. There are other people working on open office. Clearly, we want to get more people working on open office so that there's more influence and we can get some of these things fixed more quickly and a more driven project. There's a scheduling revolution. Occasionally, I've been very downbeat about this but it's really positive. We've got a scheduling right. There was something like a 20-month gap between open office 2.0.1.1 and 2.0. 20 months. Nearly two years. When we shipped, it was pretty buggy as well. It's not like we were making it perfect either. This is not really very good product development. To develop a feature and go away for 18 months and then start bug fixing it when you've entirely forgotten that it existed. There's major problems here. What we're doing instead is we're doing three-monthly releases and integrating features that are stable and tested and so on and so on and just getting features to people more quickly. That's exciting. I'm pleased about that. You see a lot more of the innovation that's going on that doesn't look like it's just the same as it always was. Nothing's getting better. We have a community in InvertCommerce that has been cleared expensively from mostly people that talk a lot. They're very articulate, right? Brilliant. But they don't actually do very much. In fact, they often exist to stop other people doing things or find someone who seems to know what they're doing and then copy what they say in a sort of demagogic way. Brilliant. There's a lot of voting. If you're on a mailing list, I'm sure none of you are guilty of this because there is this huge mail thread about something totally irrelevant and uninteresting and someone replies to this huge mail quoting the entire thing and at the bottom they go plus one. Occasionally they put their name as well and then they post this back to the list as if... I wouldn't mind if this was some really authoritative person that was doing all the work and really... but I never heard of any of them. I didn't see their check-ins. That's the kind of quotes community we have in OpenOffice and there are some ways that we're trying to fix this and get the signal up and the noise down and the work actually happening and so on but it's kind of discouraging sometimes. There's also a slightly higher barrier to entry. You need some real hardcore man's hardware. It's kind of lower since processes got faster since I wrote the slide but it does take six or eight hours to build which is fine if you're sure it's going to complete. You can go to bed or do what John was saying head out and talk to your friends or that sort of thing but it's not necessarily guaranteed that it will complete either. It often crashes quite early on and so we have a build system that tries to help this and make it build on common distributions out of the box and we put quite a lot of work into that. If you start on OpenOffice there are a number of things that Sun tends not to do. Fixing simple user interface things improving the ergonomics well, all sorts of things like that. Let me talk about what we've been doing recently. How and over time? You see that's an asynchronous unexpected question. 35 minutes ago. Mono. You see you didn't get to John Trabridge's mono slides. I hope you were on John's talk. It was very good. But we can do now beautiful mono integration to some level so you can start talking all this nice you know IDL stuff in C sharp and frankly it looks almost like any other language at this level because most of it is just you know and then you can dynamically construct spreadsheets with beautiful things and then pivot tables at the bottom as you can probably see down here. All this good stuff and do some nice dynamic data mining and so on. But there's some really interesting new areas happening here. I talked a little bit about the new XML format in Microsoft's Office 12. There's some quite revolutionary things happening in Excel there. If you look at how Microsoft is investing money inside their Office division which interests me the majority of the investment happens not in Word not in PowerPoint but in Excel. This is where the real value is. That's where the really complicated tangled numerical computation crunching happens. You know maybe you like kernel hacking and this sort of thing. That's child's play in comparison with a serious number grinding algorithms happening in cycle. You know forget the gooey piece you know there's some hard core for a peaceable computer science in that. Either way unfortunately they're very good at it. They've got some very good people on it and just to kind of show off they're increasing all their limits. So previously they had I think 64,000 rows and you know we now support 64,000 rows in Cal. We don't scale to it but in theory you can you know you can add that many rows if you want. But Excel is now up to well I don't know I think that's a million rows up from 64,000 and we've gone from 256 columns to 16,000 columns. 256 to 16,000 is a big jump and 64k to a million is a big jump and when you multiply them by each other you see that actually really there's a huge or you know three orders of magnitude difference here and so if you consider for example where Excel would use an order n log n calculation we would typically you know at least they're the best probably use an order n squared if we can you know we'll try and find a slower algorithm but order n squared versus n log n it's pretty good and when you make it you know a thousand times bigger n you soon see the problem right okay I assume you see the problem and that's assuming that we did quite well and made it only n squared often some of these things you know they like to do a list and an n squared operation in it over it and you know it just doesn't scale at all basically and you know there's some people trying hard here but the team is small there's only four people in some working on Cal. Okay there's just more in Navel there's some Intel people Google guys getting involved and in fact that's one of the interesting things about OpenOffice I worked on it for a long long time and occasionally people come to me and say look we need to hire someone you know we need someone to fix this all bit bug you created you know who can you recommend but since I started working on OpenOffice who are we going to hire you know I need to hire people you know who quick quick we need good people in the community you know that we can hire them know this stuff My guess is a revenue because you can predict what's going to happen in the future and who's doing well and who's not and what's later and then there's a massive database and they pull the data out of it and it's a big data thing because sincerely the data set is there that's currently dating maybe five minutes so we have to wait and count and that's a blow up but you know when you want to get bigger you know when the camera is decent people keep asking for pieces because it's extra price you know you use the people list for your songs and for you know restless press so exciting things there and you know some simple people looking another interesting thing since I started working on OpenOffice and that happens to me before there's two separate companies my own name or all of them they come off of the 64 bit hardware would you like a new 64 bit machine you know to work on and pull OpenOffice 64 bit and there is a little hardware floating around because it's a chicken and the next problem you know you have to be good at pulling stuff the 64 bit you can commit it but obviously you do that in hardware so anyway anyway this is a problem but for people wandering around trying to solve some of these issues case you think that all you play with boring stuff you can try to pull stuff out the team in your spreadsheet after you get a nice picture but you probably won't find it and I don't know if you think there's a soul there and quite large quite impactful exciting and interesting and after that I don't understand at all but this guy has read people on his program so we have a simple solver and it's at least something to help on that so people come and talk to me and they say to me things because I may be a small person but I'm usually hers and they say to me things why are you working on it why are you working on it and this is probably quite true but I'll show you some photographs in a minute you will play it all on someone else so that's the way it is but it turns out that there are a few ways to get rid of the one of them is to have it sitting around in the background all the time and so we've improved this because we have a nice new quick start on the sitting assist tray manages the life cycle of this application sliding in the background and I'll ask you to start that's cool, that's cool it's really loud too I'll try not to lean on the desk again good so so anyway that helps to sit around in the background and the nice thing about that is that once you've started it you can start a document very much more quickly or so you would think so we fixed that as well because that was taking about a second just to launch the thing that told the other application go load this document so we have this now tiny optimised start that Debian doesn't ship yet but will surely and that does this a lot quicker sub second, second document load you double click and it's there it's a small document obviously document load time is something we can't control because the go swallow all my memory setting in tools options that's wonderful so the real reason that I told you it got insanely technical at one point this is where it gets insanely technical hopefully you went to the Valgrin talk earlier I noticed the room was more packed then so hopefully and I need someone really bright here to tell me what the slow thing is on this drawing because I'm kind of lost which is the thing that is here and also here and here and here and here and here this is incidentally a histogram well it's more complicated now a drawing of what is taking the time during start up do you look up x? well that's part of G-Lib C and G-Lib C has this really mangled way of doing linking part of which is specified in the L-Spec hard to get around but part of which is really just an unfortunate implementation so there are some nice patches that I've been working on for some time that turn that into that which is nice that of course it's still a histogram so you don't really see the actual speed when it's still just as big as it was before but we went from 736 million simulated cycles down to 191 which is good news and we can shrink this further but it's sufficiently good news that it makes the problem someone else's fault and it's now a configuration manager that's taking all the time which is the purple bits as you can see so anyway we're working hard on this if you look at the real numbers that are actually accurate the cycle count I just used to believe as if it were a gospel and then I spoke to Julian wherever he is and he told me he sort of laughed and fell about because it's apparently only this really rough guess but this is an accurate number this is level 2 data cache misses you may know if you want to get some data it's almost always what you just wrote so you just get it from level 1 cache occasionally it's not there you go to level 2 cache and it's there it's just 100 times slur or 10 times slur level 2 cache it's 200 times slur so you know if you have a 2 gigahertz machine you can quickly turn it into a I don't know you divide 2,000 by 200 and you get 10 megahertz right so if you go around missing level 2 caches you can do some other stuff but it's like pressing that 8 megahertz button on your old machine but it's really like that and so this is what we do a whole lot off just due to some really really dumb ordering of data and searching and so we can then drastically reduce that if you look at the numbers here we have 3, 4 million level 2 cache misses down to under a million so we're improving the situation linking wise I think that takes us in real raw numbers down to unfortunately this is way slur than it was for some nasty thing with X as a spanning we're a lot quicker that's the punchline and that can take over a second of start up approaching 2 seconds from about 4 seconds of warm start so it's nearly twice as fast with the right system and if you can get the thing into G-Libs it helps which we can't because of a man called Ulrich Dripper so if you meet him, shake him by the hand and suggest you know but anyway so we've also improved the slide showing impress which is the bit that happens when you show slides and you can have seen some of the bugs earlier in Julian's talk when the colours and you couldn't see them we fixed that in mine but nice anti-aliasing smooth lines accelerated on XGL there's some nice examples here if you know before and after the irony is though there's some serious subdivision of this going on maybe you can see it maybe you can see that this guy here is a line segment there maybe there's another one there and so on and so on straight lines if you don't improve the subdivision anti-aliasing it looks beautiful so you can really clearly see these corners on where the subdivision is and we're still engaged in that so another thing that people say particularly spreadsheet focused you can see where I came from I guess VBA macros macros are in all enterprises people go and write logic in them they do funky things and I'll show you a demo of that in a second but many of the macros are really really simple actually I'll show you the demo and maybe some of the code because we paid for this code and you know I don't know what kind of programs like flow control and things like that rather than just a sort of long linear line of statements this is a this is a shameless commercial plug actually but I'll show you anyway enable the macros and what they bring this is a return on investment calculator for Nevelle in short I don't know what insure is but it's probably very good and say you click on this little widget here and that runs a little VBA thing that goes next and next you pull your company info in and you type provisioning meta directory whatever that means and then you click next and you know you flushes and flickers and set some things and selects this and this is probably quicker in excel incidentally so that you don't notice it and you can then go on and say how many users you've got and the resulting calculations and the return on investment and how complicated your organisation is and blah blah and at the end of it all it tells you what a phenomenal sum of money you'll save so thank you but this is important and the worst thing is of course that Gartner wrote it for us, Gartner consulting so it has credibility and independence something or other sorry that was back to return on investment and it tells you all this thing in the end you can print it out, this is brilliant it's sort of a tiny application in visual basic and you say wow it must be really hard to make this work well not really because when you start looking at the macros let me just go traversing down 100 of menus you see in terms of usability this is really lame it's expanded totally the wrong thing I'm actually in this Gartner spreadsheet here and you know just run this macro but anyway let me look at this most of the macros are just extremely simple and most of the macros we've analysed also still have a comment saying this was recorded by the Microsoft Excel macro recorder on blah blah blah so you know without actually doing anything really very complicated you can actually do a lot of very useful stuff for people who have these problems let me try and find some more juicy stuff go to next sheet not the most complicated function but nevertheless very useful if you don't have a web browser and a remote thing so you get the idea good so this is a very useful tool and there are some more funky demos that are less obviously partisan let me find one of them here and you can achieve a lot let me encourage you to go and write VBA macros in open office and if you come to my tutorial later I'll show you quite how much better VBA is than Starbase and there are some quite staggering differences and some quite amusing weirdnesses as well incidentally you have to explicitly enable the macro otherwise you have real problems hyper cycloid you can have your favourites that can be generated some simple charting some simple VBA and a pretty result comes out of it and there's even a what is hyper cycloid which I never knew apparently you can make it with several spirograph circles and something else so it's really bad it's not even 20% of the functionality is implemented and yet we can do a lot of useful things with it but of course maybe your 20% is different to ours so I'd encourage you if you have access to VBA macros send them to us run it and see what doesn't work and we'll try and make it better so cool things in open office 2.0x so one of the cool things is as I said earlier hopefully some of these features will come in the stable when they're ready ready for prime time and we'll have a more manageable codebase 64 bit work, there's a whole load of ongoing work there and the Val and Intel and so on and we have binaries running and we have a game in calc this is one of the first things actually it's a shooter moment so it's getting there but there's still lots of testing and little bugs and help needed they're going G streamer support media is a bit of an issue they've ripped the whole media thing out and used the java media framework and the reason incidentally son does this they get bad press for using java but if you're a small team of only 150 programmers or whatever either when you need to play a video sit down and write an entirely new cross-platform abstract video player this is good and then license all codex and blah blah blah or you can say our company already has a division that does this over there why don't we blame them when it doesn't work and use the tiny bit of code to use the java media framework and it's entirely someone else's problem this is a very attractive thing you can understand there's no evil malice let's lock out free software here there's just a oh dear I don't want to disappear down a huge black hole here but unfortunately we then have to go in filling the holes in afterwards so we're going to need G streamer support and some multimedia stuff there not impossible to do similarly accessibility is done via java bridge so it works on linux and also windows unfortunately it works on both rather slowly so there's a native bridge there that actually we're probably shipping now to improve accessibility and make it much more responsive and easy to debug and deploy on free systems Microsoft access import we have actually got an access back end that you can talk to your access database and you can load your data out of it and put it in your spreadsheet and do a massive pivot table on it or whatever the problem is that the code is very good at what people test it for doing which is essentially migrating away from access use mdb tools brilliant dump your entire database abandon access databases and wander away unfortunately the way our database thing works you have to use their implementation of all the querying features and inner joins it's a noddy noddy database in that sense but it's really good at getting data out so we need to refactor this and turn it into a sort of importer pull your data in and allow you to shove it somewhere else and then still access it along with your templates and forms so there's some chunk of work going on there layout, the layout looks awful so there's more work happening here to use either something like Zool I thought Zool was really cool for using this splitting the code out as well as the actual layout very powerful into the dialogues, the dialogue runs itself the dialogue is a program in its own right that you just say go do whatever you do and it throws up the dialogue and you can edit it all and it does seem good but the problem is you have to rewrite most of the code in javascript and then I talked to some Microsoft people and they said yeah actually we tried that for like three versions ago and it works really well for the search dialogue or something but when you get to complicated it's a nightmare and low we stop doing it but it's an interesting task improving the look of it is worth doing and chiro rendering throughout the application not just a slideshow this whole lot of stuff that needs doing there so how am I doing for time you haven't shown me any pictures yet 17 minutes so 2.0 is a massive improvement it really is amazing it's incredibly powerful, great feature depth it's an amazing sort of spearhead of free software on proprietary desktops out there lots of people are using it lots of people want to start using it the sluggishness and the bloat that you perceive particularly on Unix but it's not so much present there on windows is on the decrease and we're beginning to win some of these battles which is nice so I'm optimistic that in two years time I can come back it'll start up like that, you know? faster than you have to and OpenOffice is a really good strategic place to make your blow for freedom on that basis there's lots of things to get involved with there's a wiki to read but you have to sign the JCA before you do that and my tutorial shortly afterwards I encourage you to come and sign the JCA if you're interested so you can contribute and get involved so thank you for being very patient and kind I meant to mention that if there are questions as I hope there will be a few we have these relatively bygone era with zimeon monkeys on them which are frispies and we also have some open suzer shirts of gigantic proportions so we're trying to get rid of so on that basis hopefully there will now be no questions but if you have a question, please let Rip you've had one already so let's wait there's probably some, you know you always programme at times that are kind of scared of speaking in public, you know if you feel yourself wanting to question something yeah there's a man in the back, hooray any chance of integrated version control yeah that's a really good question so let me, I might kill someone with this so you know oh it didn't get there, if you could help it on its way that would be great so there are a whole load of things built into open office that make it really nice for reuse and kind of version control things so compare changes so you can get your two documents side by side and see what happens there's change tracking inside the document so you can look at various different revisions and if you change what and accept it and so on there's a lot of funky work there in terms of integrating in version control or content management systems there's also some nice hooks so you can define your own stream there's a virtual filing system of its own invention of course and it tells you when documents are open to closed, renamed, resaved and so on all the hooks you need to integrate it with the content management system and you know add commit messages and annotations and blah blah blah so yes it's easy enough to do and we should do more work on that we're doing some work in fact there's a guy called tall little fist there yeah and he's doing some integration with content management type systems using ODMA so yeah any other questions this man here yeah not a lot around well you saw the box, the chew your entire system memory box there yeah you probably don't want to have that checked but you know for thin clients the question is thin clients how do you have it working the nice thing is the open office looks incredibly bloated and huge and to be honest there is a lot of code there there's you know 50 megabytes 55 megabytes of dot text just raw output that's before you add the data and the symbols but all of that is of course well the vast majority of that is shared so a thin client system actually start up and a memory usage should be substantially better than on a single client system because you share all of that huge chunk of code most of what makes open office pain to start up so yes good the runtime allocated memory usage is about you know 4 meg for an empty document now of course you know it can be improved but you know no real tips there on doing that I think it's pretty good sun actually running themselves on of course sun ray so you know they've done a chunk of work there themselves to make it work nicely okay this man here oh what the heck okay a lot of exchange of documents of Microsoft documents with open office documents and well we seem to have we find some problems and we really find styles for the formatting and we exchange them back from office and back to open office then we get a little bit more can I be a bit mean just briefly to you will you co, that's good, I just give you a frisbee okay so Mae'n gweithio'n dyn ar gyfer yng Nghymru, i ddwy'n gweithio'n ddweud. Felly, rydyn ni'n gweithio'n gweithio i ddweud. Mae'r ddweud yn amlwg. Felly, mae'r ddweud yn ddweud. Felly, rydyn ni'n ddweud? Yr ysbyt yw'r ddweud i gynnig yma. Mae'n ddweud i'r ddweud i'r ddweud. Mae'n ddweud i'r ddweud i'r ddweud. Yn ymgyrch ar ychydig, mae'n ddweud i'r ddweud i'r ddweud. Many bó Things that complex and pushing attack really does have to be a better way to...into침 Psalms evenings. Yeah so it probably is a better way, but the thing is, from my perspective as a Novella obviously we want people's data to continue working and provide a solution for people so that is primarily where we come from. So yes we tend not to do let's completely invent the whole document processing paradigm thing, and if you look at the successful attempts to hey man let's completely reinvent the world, there are very few, far between. Particularly in this area, however there is some pretty funky innovation gathered is hear the apple's key not it's amazing presentation effects are there some really new stuff but my feeling would always be to be on incrementally go on what you have don't start from nothing re-invent compound system media system build on what's that then improve it because a lot of and if you've used OpenOffice a lot of your pain is probably minor ergonomic stupidity rather than you know like where's the styles why can't i find these styles why are they not docked on the right hand side gallwch chi'n bryd i'r rhaglen a gwybod iddyn nhw? Mae'r rhaglen ystynnos ffordd cydnod? Mae'r rhaglen ystynnos ffordd. Mae'r rhaglen ystynnos ffordd oherwydd oherwydd, os ydych chi'n bryd yn gwybod i'n blynedd. Ondo, mae'n gwybod ystynnos ffordd i'r rhaglen ystynnos i'r rhaglen i ystynnos i ddigon i'r pandogau. How long have we got? Ten minutes. This is good. You go. You showed us some macros from the offices. Security about macros. Macros security. Then he dealt with your macros, but most of the time macros don't do anything except calculations. Absolutely right. So the question is, is it good enough just asking people if they want to enable macros? The answer is probably not. There's a tiny feature that he's writing here. You can scan along a string. You can write this feature. But it turns out that when you load macros in, particularly this kind of macros, most of them are just non-macros. Someone just opened the macro recorder once, that kind of thing. It's stored some attribute and there's actually nothing there. It would be great to filter out those false positives so people don't come hard into just enabling them. The second thing is, at least with StarBasic, the primitives are not really there to allow you to do some of the things that are possible in Microsoft's Office Suite. Hopefully we can do a better job of sandboxing them and containing them and stopping some of the most grotesquely stupid things that can happen. Of course, Python is quite a powerful language and if you can run arbitrary Python stuff in your macro theory, you can do immense damage. The other thing is lockdown and management. I'll talk a bit about that in my tutorial. How can you stop people being able to run macros at all, maybe across your organisation, or at least make it very difficult for them to shoot themselves in the foot? It's not a very satisfying solution, but I think it's better than what Microsoft can offer. Sandbox, absolutely. Why not? Absolutely. I think partly we do that by, if you use StarBasic and Java, you're fine broadly because they don't allow you to do some of the stupidest things. But if you start using Python, it's powerful for a reason. People like to do powerful things. As you can see, you can do some very useful things with macros. If you turn them all off, you read the world of what is actually quite a nice, very rapid application, very simple to use development environment, for simple numerical things. You have a T-shirt. There you go. Gentleman in blue. I think if you're a baseball player you have to also do this. Don't you sort of frequently in a rather strange way. I'm not good at that. Are we planning to address the usability issues? Yeah. I think probably the best way to do this is start a mailing list and get everyone to talk on it for weeks and come up with some ideas and mock-ups and argue about them. Are you going to help me fix the usability? Fair enough. This is the problem with usability then, because there are lots of people that want it, and a few are people that actually create it and are prepared to do the fascist things necessary to get it. So yes, I'd love to improve usability. Yes, there are lots of fairly simple things that we can do. Yes, we've done some of them, but there's plenty of scope for work there. Sure, and I apologise for it. It makes free software look bad to a lot of people. So to what extent can we decode the binary Microsoft Office file format? Well of course we stop at Word PowerPoint Excel, right? We don't do project planner and Vizio and some of the other more interesting eclectic acquisitions that have been made and have quite different file formats of different structures and so on. How much have we figured out of these? A huge amount, actually, amazingly. I think what you see in OpenOffice is probably the world's best outside Microsoft import and export across the board like that. It's really extraordinary. You can go to SoftMaker and you can see the three features that fail to implement that they've pulled out and they have implemented it. And look at the screenshots and it looks terrible. It looks like we're failing all over the board. It's convincing for me. I look at it and I think, what? How come we're so bad and they're so good? But actually, then you download their free trial version and you're like, what? This is useless. It doesn't even do XYZ and so on. So I think, sure, we do quite a good job, a really quite a good job, but it's a difficult problem. It's a really difficult problem. So on my system, I haven't had proprietary, unfortunately, licensed fonts from Ag for Monotype. And that means when I load my document, it lays out broadly if I use Arial, Times New Romans and Courier New, it lays out like it lays out on Windows. The problem is, of course, if you don't have those fonts, it lays out radically differently. And the whole structure of your document can often depend on proprietary fonts. So I'd encourage you, if you're a free software type person, to try and use fonts that are free. Sounds silly, doesn't it? Say you write this long screen as MI5, and you've got two pages and you say, kill that bloke there in the text. And as you lay it out, your points are this guy, and then you load it in an open office and now the points are this guy, who's actually your contact. It's not so good. And in terms of document, archival, and long-term storage and retrieval, it's a bit of a nightmare. And the worst thing is that this is a huge IPR issue that we can't really discuss, and it's unsolvable, basically. There is no solution. You just have to use free fonts everywhere. And this is how Microsoft interoperability works. They have Arial everywhere. They have Times New Roman everywhere. It's the same font, the same metrics, and that's why it works. But we don't have that necessarily. And yeah, you can use some funky web download, cache web font thing. But I don't understand to try and overcome that, but it's a serious problem. So yeah, that's my hobby force. Two more minutes, two more questions. This man here and this man here. You first. Why won't Ulrich Dripper fix my G-Lib C problem? Well, he doesn't like to talk to me for reasons entirely unclear. He thinks there's this thing called prelink that solves the problem. Now, prelink does attempt to solve this problem. It's true. But what prelink does is essentially fragments your disk so that it may be faster to link, but you now have to do hundreds of seeks all over the place to load your data so it's actually potentially slower. And it also takes a huge amount of CPU time on every end-users machine. It needs rerunning every time you upgrade any library. And it doesn't work for DL-opened libraries, which includes the vast majority of open office. However, there are some minor theoretical efficiency wins to it. For example, if you're running multiple programs and they are prelinked and there are not very many fix ups, then you can share some more pages between them. Of course it's meaningless for open office. And it's quite possible that, you know, direct linking and some of these other hacks would have some minor slowdown on the incredibly elite hyper-optimised, you know, there's a lot of hyper-optimisation of the code pass that do this currently rather silly algorithm that make it rather nasty to refactoring. There's just a lack of, a total lack of actually getting this fixed. I mean, or it just doesn't want to engage at all. And what's that? What stops you from forking is a real pain to maintain a fork. I happen to maintain a semi fork, a small branch of open office. And it's a nightmare. Maintaining a branch of G-Lib C, particularly in a very hyper-optimised elite piece of code, and refactoring it as patches, especially when there are several features that you might want to merge separately in an optimised hash values, optimised direct linking. You know, it becomes depressing. So, okay, my secret strategy is this. A nice set of patches that make linking much faster. We'll get them in SUSE, we'll get them in GEN2, we're already doing some research on doing this and providing some valuable test data. And, you know, we'll try and encourage distributions to ship it, and in the end, reason will prevail. But, unfortunately, it's a rather distressing and you can't do as good a job as you do if you get upstream because you can't refactor as much. Because patches suck. Okay. Did I have one more question? It's a quick question. Who was it? Yeah, quick. Why do we suck, yeah? Yeah, no, that's a very good question. So, the question is why do we suck with PDF export in OpenOffice 2.0, and I have no idea. I'm sorry. There is a new PDF export dialogue as you saw there and you can check all sorts of options. Your question may be the flipside of why does my image look so bad when I print it? You know, so we've raised the, you know, the resolution of the images. So, I don't know. We can look into it. Thank you very much.