He heard that sleeping with big guys was the only way to succeed, he didn't believe it.
The fact that there was no more twin beds room in the hotel was unexpected…
Introducing the innocent Ploum and the Nasty Desrt in a new Mountain View based horror comedy :
Soon in the theater : In bed with Ryan Lortie II, featuring not-so-innocent Vuntz and still-naughty Desrt in Dallas.
]]>Back in 2005-2006, my good friend Patrick Rácz developed a small pager for the GNOME desktop. Since then, I've been an happy user of this little application, recompiling it with every Ubuntu release. In fact, I just cannot live without it anymore.
We thought that it might be helpful to other and, for years, we said : "we should publish it". Today, after fixing some small bugs, I'm proud to announce the release of Gpager 0.3 "Scenes From a Memory - Finally Free". I created a Launchpad project and even an Ohloh page.
Using Gpager is really simple :
Pager properties have to be set independently in gconf or by adding a pager to a panel and going in its properties menu (not really handy, I admit).
Dedicating already a lot of time to Getting Things GNOME!, I do not have any plan to further develop Gpager. But I will gladly merge patches or solve small bugs that might be reported.
Testing gpager from trunk is a simple matter of :
Alternatively, you can download 0.3 tarball.
I'm open to any suggestion and I really hope to see something similar to gpager in gnome-shell.
Packages for Ubuntu Karmic are available in the PPA.
]]>Nevertheless, I don't have a Wikipedia account. I don't see the need for it. Sure, it could be useful for others, but I don't think I'm really good at writing encyclopedia stuffs, I don't find it funny at all and I believe that I'm already doing some useful stuffs. I don't want yet another account, I don't want a password nor statistics.
Of course, I was correcting mistakes when reading them, I was sometimes adding one or two sentences because my knowledge of the page's subject was enough. I even created a few "stubs" when I thought it was missing. Some of them became real well-formatted Wikipedia pages afterward.
It was back in those days…
I don't do that anymore.
For the last year, every change I did on Wikipedia was a catastrophic failure. As I was not logged in, they were most of the time seen as non relevant. Most of my changes I manually followed where flagged with the infamous "Reference or Citation Needed", even those that could be confirmed by any history book.
Sometimes, you know something so well that you can't even tell a source for it. Ask any Belgian the name of the first Belgian king. And then a reliable source to prove their answer.
Anyway, if I replied with a website source, it was rejected because no blogs or forums are allowed as a reliable source. Think about it. A blog is simply a website with chronological content. Most of modern website with information and content are blogs. ArsTechnica and Slashdot are considered as blogs. The fact that a date is put on the content should not interfere with the quality of the said content. In theory.
Wait, it's worst!
When you edit a page, even for a one-letter typo change, the page immediately falls in the whiter-than-white-wikipedia-ayatollah-commando-squadron radar. And, oh surprise, they discovered that the page hosted for years a content that was not perfectly following the Wikipedia religion. And, as every good zealot should do, they immediately corrected the error, regardless of the fact that the content was fine and useful for years.
On one article I was reading about some sport, I corrected a mistake in an URL in the external link section. One day later, the whole section was removed with all the links because every pointed website was considered as "a blog". Those links were really relevant to the subject and very useful for me. But they were "blog". Now, I consult the history of that page to have the links.
Another article[1] I recently corrected for a spelling mistake was deleted as "non-notable enough". The fact that I was looking for that article, that I found it, that it was existing for weeks or months and that I corrected a mistake should be a proof that it was at least an useful article.
But Wikipedia doesn't want to be useful anymore.
It's a common story in the human species. First, we want to achieve a goal. Second, we discover that we are all different[2] and that we need some rules to organize our work. Third, we make the rules really complicated to fit every corner case. Fourth, we completely forget the goal of those rules and we apply them blindly for the sake of it. Fifth, we punish or kill those who don't follow the rules as strictly as we do.
Every religion started that way, most companies internal rules have the same origin. Sadly, Wikipedia is no different. For some time, I hoped but… no. Sad. Let's pay extra attention that my projects and my own life don't follow the same scenario.
]]>
3 years ago, I was a SoC student and developed the now abandoned Conseil but I learned a lot from that experience.
This year, one of the GNOME SoC projects was related to Getting Things GNOME!, the software I started with Bertrand. The project was to add the concept of geolocalization to your tasks list so you would be able to see where you can do tasks. The candidate, Paulo Cabido, seemed bright and skilled. I was the mentor. Strange to be on the other side of the fence.
Let's kill the suspense : the result is more than a success. It's wonderful. I'm proud to say that GTG has now a whole plugin engine, a geolocalization plugin using libchamplain and geoclue has a python "extended-binding". More : GTG has now one new and very talented core developer, the GNOME community has one more bright member. Rock-n-roll !
Seeing the result, I might say that I was a good mentor. But I was lazy and I didn't worked 10 hours a day for that SoC. Let me share my little recipe.
The student : find a very talented an creative student, with technical skills. Be sure that the student is actually skilled enough to do basics stuffs. Write the proposition with him to be sure you have the same understanding. 3 months is too short to learn and master a new language or a new programming paradigm, at least for most of us. Paulo demonstrated his Python skills and even sent a small patch to libchamplain, showing that he didn't feared C, bugzilla and other tools. We discussed a lot together to write his proposition.
Communication : reply to every email of your student. Even a bit late, even with a short "I don't know" or "No time, sorry". Reply. No excuse. Reply. If no news and no sign of activity during a few days, send him an email. Most of my communication with Paulo were one sentence emails. Quick and efficient. When we wanted to discuss something, we fixed a meeting hour to discuss on IRC or Jabber. Frequently at the beginning, less at the end.
Freedom : your student is *not* a coding hand. My hidden agenda was to make him a core developer of GTG in 3 months. The project itself was, in my mind, only an excuse, a learning tool. It means that your should never impose anything (technically speaking). Respect his choices and, if you disagree, that's a very good sign. Discuss to find a solution and to understand his view. Consider him as a coworker and let him make his own choice. You are doing it right when your student needs you less and less. In the last weeks, nearly everything I was doing was reading his commit logs and his code.
Resource : it's his project. Not yours. You are only a resource that can give him informations. He will makes his own choice with those informations. Sometimes, let him think by himself. Just answer "What do you think ?". When you don't have an information, tell him how you would find this information but don't do it yourself. Paulo quickly grasped that and I believe he was efficient to use me as a resource, not as a supervisor.
Community : Introduce him to the community. Tell him to ask this developer or to see on this mailing-list. Bertrand and the whole GTG community were in fact co-mentor of the whole SoC, thanks a lot to them. I remember the first time someone asked me a question about GTG and I replied "Ask Paulo, he's in charge for that". I shivered, it was so nice to feel, on that very second, that he was my co-developer, not anymore an anonymous student.
Honesty : That's maybe the most important point. Be straightforward and ask your student to do the same. Never hide problems. It is perfectly OK to say "I didn't work at all this week" or "I screwed my branch". We don't care about excuses. We only want facts. Everybody does mistakes and has problems. But if we communicate about it, we can solve everything. It also helps a lot to be honest when fighting against procrastination. If you are direct, the student will be. And never be angry at a student : you are the mentor. A lot of student fail because of bad mentor. Once or twice, I was a bit annoyed by the way Paulo was working. He was not listening to me! Or maybe, were my explanations not so clear? Taking the time to explain one more time, from A to Z, solved the problem. I was not happy with what the student was doing but, in fact, it was my fault!
This was my first experience as a mentor and I tried to stick to those rules. As this was a wonderful experience, I thought that I would share them with you. Of course, there are hundred of successful SoC. This is not *the* way, this was my way. Some will say that only the first rule was really important, and that Paulo would have succeeded even with an IRC bot as a mentor. And you know what? I believe they are right!
Well, forget that! Each organization is awarded money by Google to send two mentors to the mentor summit. If I want to be chosen amongst the GNOME mentors, I should probably say that I worked so hard, with sleepless night of code and student coaching. Hey GNOME foundation, did you hear? I worked very hard and really think that I will be a good representative of GNOME in the mentor summit and I will send a postcard to every member of the board of directors[1]. ;-)
Enough speaking about myself : Congratulations Paulo, welcome to the GNOME jungle. I hope that you and Bertrand will soon be GNOME members and that the whole GTG community will continue to rock as it is right now.
[1] At least I can try, can't I?
Just think about a machine that will need input. Add lot of buttons, more than one hundred, nearly all of the same size and color. Spread those buttons in a total random order, without even aligning them. That's it, you have the most widely used human-machine input mechanism : the keyboard !
Scary, isn't it ?
I used to hate my keyboard with passion. Whatever I was doing, I was always pushing the wrong key. Most of the time, I had my finger between two of those misaligned buttons, pressing both for the same price.
That's until I discovered the Typematrix keyboard. Since early May 2009, everything I've written was on a Typematrix. From coding for Getting Things GNOME! to ranting about usability. And I must say : it's just too good to be true. Yes, I love this keyboard.
I also took the oportunity to learn the Bépo layout, which is the french speaking variant of the Dvorak keyboard. Indeed, typematrix is selling them with a Bépo skin and thus I've only used my typematrix in Bépo.
This is not a sponsored post but when I like something, I want to share it. I also think that promoting a company might help them in the long run and thus I will be able to buy more marvelous keyboards in the future.
So, why is it so good?
1) Columns are just perfectly natural. It's just working, your brain has a lot less to do. It's so natural than people don't even notice it (after all, most small keyboard on electronic devices are aligned and nobody complain about it).
2) The enter key in the middle is just perfect once you are used to it.
3) Multimedia keys working out of the box and easily accessible.
4) Compact, light and easy to travel with or to put above a laptop keyboard.
5) Pure post-stalinian look. Yes, I like that, I never liked glossy roundy Apple-ish stuffs (but I admit that having a wireless diNovo Edge-like Typematrix would be the hypish thing on earth).
6) Finally, a compact keyboard with an usable numpad !
7) Arrows and home/end keys do make sense, really.
8) The skin add a very soft touch sensation. It also reduce the noise greatly. Note the the skin is optional and that the feeling of the naked keyboard is very nice too.
9) Very comfortable strength of the keys feedback. It's not a cheap keyboard and your fingers will thank you.
10) I remapped the "desktop" key to launching a terminal. It's the brightest idea I had in the whole 2009, it's just so nice, how was I doing before ?
11) Alt key is *very* accessible, making it really easy to access the third level.
12) It's very pleasant to have it on your knee when working without desk. It also saves your future generations from the CPU heat.
13) Switching to typing with 10 fingers on a Bépo Typematrix totally relieved me from my hand and fingers pain.
Well, of course there are some drawbacks.
1) You will curse every other keyboard you have to work with, really.
2) Copy/Cut/Paste keys are not working in every application (but most of them are fine).
3) The dark skin is a dust collector. It's completely disgusting after a few days. Hopefully, you can easily watch the skin under the tap. I highly recommend the blind keyboard with a transparent skin (I have both and that's my preferred one).
For geeks spending work, leisure and even sleep hours typing on a keyboard, it's just as indispensable as a proton pack to Bill Murray.
And then you realize that the size factor, the form factor, everything makes it a perfect laptop/notebook keyboard.
Mockup done by blogeee
I found the keyboard of my dreams. Put it in this netbook and this would be the computer of my extatic bliss…
]]>
I mean, seriously, have you ever think about how completely stupid is the idea of a "tray icon"? Can you imagine how black magic it should be for a new computer user?
Windows 95 came with a new concept called "taskbar" when you could have a list of opened windows, a start button and a notification area. Due to an historical misunderstanding, some people started to call the notification area by the name "systray". This was wrong but this misconception might have helped to forget the original goal of the notification area.
As computer performances raised, people had more and more applications opened at the same time. Unfortunately, the taskbar list of opened windows was limited by the width of the screen. In the Unix-like world, the solution was found with virtual desktops. But there were no virtual desktops in Windows.
Developers then found a workaround : abusing the notification area and putting their application there instead of the list of opened windows because notification icons were smaller. It was clearly an abuse but it had many benefits for the developers. One of those was that the user would be less likely to close the application, unaware that it's running. It allows the application to have more control on what the user is doing, to start faster, to continue to do some processing in the background. Remember : we are in a commercial world where each application want to be seen as much as possible by the user, using non-standard skins, splashscreens. We are in a world where we don't want to help the user, we just want to be familiar to the user and to look completely different so it worth the price to spend.
With Windows XP, Microsoft acknowledged the abuse of the notification area and offered a function to "hide" some icons from the notification area. I find that really funny. I mean, imagine that you buy a brand new car and that some mysterious indicators started blinking everywhere on your dashboard. What would you think if the car maker, knowing the problem, offered you a pair of sunglasses with the next model of the car?
Anyway, this "feature" reinforced the idea that it was a "systray", something not well defined but where you could put icons if you wanted them to be smaller than in the tasklist.
The notification area of most Windows system is now fucked beyond any repair. I highly recommend the euthanasia of the whole thing.
I've seen with my own eyes, a fresh OEM installation of Windows XP on a Toshiba laptop starting for the first time with 43 (forty-three!) icon in the "tray"… by default! The notification area was starting before the middle of the screen[1].
At this point, you might think : "ok but what's wrong with putting icon in the systray?". The answer is in three words : consistency, logic and usability.
You might not realize it because you are using tray icons for years but there's no logic behind the systray. Once a window is open, there's absolutely *no* way of knowing what will be the result of the close button. Will it close the application, stoping its current work or will it just hide itself in the tray?
Even worst : the systray is sometimes called on close, sometimes on minimize. And when it's on close, this is not trivial to really quit the application.
Using the notification area as a systray break the window manager !
So, only because you, application programmer, want your application to behave in your non-standard way, you break the usability of one key component of all your users! Yes, it is that bad : if your application use a tray icon by default, you are breaking completely the usability of your user's desktop. Even if it's only your application. It is proven that when beginner met an exception to the rule, they don't trust the rule anymore, seeing once again the computer as a scary stuff without any kind of logic.
For beginners, it's a nightmare. There's no way to know what an application will do when you press close. There's no intuition. Marie was quick to grasp the taskbar concept but she still doesn't understand that some applications are not in the window list. I pointed her the notification area but she said to me that it was so tiny that she doesn't want to understand it. Jean, on his side, is still fighting with the taskbar concept. He doesn't understand the difference between closing and minimizing[2], I don't believe I will be able to explain him the so-called "systray".
On the other hand, application that behaves normally let the user choose to put them in the taskbar or on another desktop. User is put back in control! Yummy!
So much for usability? Wait, there's more. Using the notification area as a tray make it useless. When there's more than 4 or 5 icons, the eyes simply don't catch any new icon anymore. Finding the icon you want is slower which each icon you add. Of course, if your application was the only one to use the notification area as a tray, it would be fine. Unfortunately, it is not the case.
Just one real life example under GNOME:
- you are connected on a wifi : one icon (it can be seen as a notification)
- you are on battery : one icon (it's an important notification)
- you have taskcoach started on another desktop : one icon (or just try gtg)
- you start tomboy from the menu to see a note and don't have it in your panel : one more icon
- you have lot of ram and want OpenOffice to start faster : one f*** completely useless icon
- you listen to music with rhythmbox which is on its own desktop and controlled through music-applet : nevertheless one icon more[3]
- you want to take regular breaks and use workrave with the applet. You still have this cute but useless icon.
- you are connected with Empathy on Jabber. You have no new messages but there is still one icon, redundant with the User Change applet.
This is a basic session that can be handled very well with a few virtual desktops [4]. You can see that there's nothing fancy (it's not far from my own default session) and that I only took icons that cannot be disabled (I disabled tracker for example). It means that, by default, the notification area is already 8 icons wide and only 2 of them are real notifications!
With 8 icons, you already take a lot of space on smaller resolutions like notebooks. It also means that there's very few chances that the user see a 9th notification icon unless it's blinking. And now the war to get user attention start : Ubuntu already abandoned the too crowded notification area to start the update manager automatically. According to user complaints…hem, feedbacks, this is one idea which is so bad that you can put it with the idea of using the notification area as a second taskbar. Bad ideas call bad ideas and, eventually, you end up with something as unusable as Windows XP.
Remember : it's not because you are the developer of an application that you can choose what your users desktops will look like. You have to respect your users and the first thing is, by default, to behave like any other application. Your application is *not* special.
Good applications are applications we don't even notice.
One argument that I often hear is "systray works also on KDE and XFCE, it's easier than an applet. Applet are old and complicated".
I agree with the statement above but I have two questions :
1) Do you really need that icon? It's not because it's easy to do that you have to do it. It looks like all new application quickly gain a tray icon only because it's cool.
2) Do you really believe that because our underlying implementation sucks, we have to decrease usability, user experience and perceived quality of our desktop?
Do you really believe that user should pay the hard price for our laziness?
If we don't do anything, we will end in a few years/months with a default 43 icons wide notification area. Fortunately, there's hope, you can too correct your own application or report bugs against misbehaviour in your favourites applications.
I know that you are so used to this systray behaviour that we cannot just get rid of it. We also have to respect historical power users. So, what should we do?
1) By default, no application should have a permanent place in the notification area. If you still want to have this behaviour, make it a plugin or an option.
2) By default, your application should close on close and minimize on minimize. It makes sense, isn't it? Hiding in the tray cannot be a default behaviour in any circumstances. There's of course nothing wrong to have it as an option as it's then the user choice, not the developer's one.
3) Think twice before having an icon in the notification area. Kill useless icon (like openoffice starter). And rethink twice about adding that icon.
4) The icon should indicate an event or something new in your system. Pidgin and Specto already allow you to have this behaviour. They don't show in your notification area unless there's something you should check. You cannot believe how good it is to see the pidgin icon only when I have a message. I also rarely miss a message, which was not the case previously and, because it's not blinking, I can choose to ignore it for some time if I want to. More freedom and power to the user = win[5].
5) If I have the applet, shouldn't be obvious that I don't want the icon in the notification area?
Not too hard too do, isn't it? So, let's make the desktop more usable, one icon at a time. Yes, I like the smell of burned systray in the morning!
Part 1 : Introduction
Part 2 : Home is Desktop
Part 3 : There's no tray icon in GNOME !
[1] I should have taken a screenshot, I'm sure you don't believe me
[2] I will talk about that another time
[3] should be solved in the next release, yeah, thanks rhythmbox developers!
[4] Maries seems to like Virtual Desktops a lot, she asked me how to increase the number up to 6
[5] that feature alone is the reason why I can't use Empathy at all
Amongst the advantages of following this specification, I list :
Your application should not have its own folder anymore (and should not use another software hidden folder like .gnome2).
User data should go into $XDG_DATA_HOME (which default to .local/share), user preferences should go into $XDG_CONFIG_HOME (which default to .config) and cached data should go to $XDG_CACHE_HOME (which default to .cache).
Of course, there's no need to read the environment variable yourself : most language provide a library to use XDG folder. For example a patch for a GTK application or one for a python application.
There's a little trick : just imagine that it was deleted. What would you think as a user ?
If you would cry, running frenetically to your backup, it means that it belongs to XDG_DATA_HOME.
If you would think "Damn, I will have to reconfigure all", it belongs to $XDG_CONFIG_HOME. This includes user installed plugins even if those plugins might themselves have files in XDG_DATA_HOME. The exception I see is the usernames/passwords for external services. Those are clearly configuration stuffs but, imho, just fit better in XDG_DATA_HOME.
If you would just think "It's bloody slow those days" then it is obviously part of XDG_CACHE_HOME.
Of course, I'm sure there is some corner cases. But take the time to choose carefully instead of just putting everything in XDG_DATA_HOME, defeating the purpose of this specification.
A lot of applications are already following this specification (VLC, Wormux, Totem, Metacity, Cheese, Deskbar, Pyroom, Getting Things Gnome) or will soon (Rhythmbox has it in SVN, GStreamer will have it for 0.11,..)
GNOME projects are listed in the tracking bug and you can help to complete the Gnome goal proposition.
And you ? Is your application already XDG compliant ?
PS : If you are not the maintener of any applications, sending a patch to your favourite project might be a great contribution. If you never contributed code to any project, take this opportunity : writing a XDG patch is really easy to do and it could be your first step to world domination.
]]>
Bertrand and I are very proud to announce you the first release of Getting Thing Gnome!, a personal organizer and todo list manager for the GNOME desktop.
Nearly a year ago, we started discussing about softwares to keep our todo list and to be used with our more or less Getting Things Done methodology. We wanted a GNOME software, with a very simple interface, with the ability to quickly add a task without filling a huge form of hunderd fields. Nothing was nearly what we wanted so we decided to start our own business. In October 2008, we met for a brainstorm meeting where we decided the basis of our future interface.
A bit more than 4 sleepless months later, we are proud to offer you GTG 0.1 "Just 5 minutes more". This release is dedicated to the countless times I replied "Just 5 minutes more" when I was coding and my beloved one was waiting for me.
GTG is still in its early stage and an experimental software. This is more a working proof-of-concept than a real software. We are really curious about every feedback. Don't hesitate to send us your ideas and report bugs. We don't want to keep GTG as our own project. We hope to see it, maybe one day, becoming a component of the GNOME desktop and, in this regard, user feedback is really important to us.
There are a few things that were on the planning but we couldn't achieve for 0.1. The most important one is probably the ability to store your tasks on a server. In this regard, we developed an extensible backend system for GTG. It should be easy for any python coder to write a backend for GTG but currently we have no GUI for allowing the user to change his backend and we have only one implemented backend : localfile.
Amongst other stuff we didn't achieve : translation infrastructure using launchpad (we simply don't know how to do it properly) and unit tests (because, after all, it was an experiment, nothing else). But we hope to bring you all that goodness for GTG 0.2 with ambitious stuffs like a gnome applet, a full Dbus API, Gnome DO plugin and many others.
So, grab GTG on our PPA, install it and send your feedback. And if you like GTG, join us on our user mailing list, on Ohloh, on #gtg (GimpNet) or anywhere else. We are looking for all contributions : documentation, translation (when it will be ready), a new icon, ...
I hope you will be as enthusiast about GTG as I am. I'm using it all the time, at work and at home and it changed my life. That's why I want to publicly hug Bertrand for all the hard work he putted in GTG. Without him, I would have never the motivation and the ideas to work on this project. He did all the boring stuff in GTG that I would have never done and it's for me a real pleasure to work with him. Please kudo him on Ohlho, he deserves it !
PS : if you want to be productive and get things done, here's my advice : never develop your own GTD application. Never !
In theory, the apt-get update/upgrade works well with the update-manager interface.
From my experience, people have a lot of difficulties to tell the difference between a normal upgrade and a full upgrade (dist-upgrade). They don't really grasp the concept of "new version". I don't believe users have to learn that concept but it has some consequences
Upgrading is way too slow. It takes ages. Why could it not start to upgrade a part of the system as soon as some packages are already downloaded ? Also, there's very few feedback that the system is actually doing something.
You cannot just let your system upgrading because it asks you some question and will wait for the response. Why cannot those questions be asked before beginning the upgrade ? Oh and, by the way, none of those questions are understandable by normal users. And most questions are about config files I never touched.
Upgrade just crash once in a while. It knows that a simple apt-get -f install will solve the system but insists that you do it yourself. Also, sometimes it looks like it is finished but launching apt-get dist-upgrade once again shows more than 300 non upgraded packages with are in an unknown state. In the end, upgrading quickly become a serie of apt-get dist-upgrade, apt-get -f install, apt-get dist-upgrade, apt-get -f install until no more packages left.
Shutting off the computer in the middle of the upgrade process leaves it, most of the time, in an awful state and you can be sure that X doesn't work.
After upgrading, a lot of old crap stays on the computer. You have to run apt-get autoremove --purge `deborphan` multiple times but it's not really enough because the program is not detected by deborphan. Just remember all the old cupsys stuffs, gnome-volume-manager, things like that. On most upgraded computers, they are still there !
Problem 1,2 and 5 imply that geeks will just reinstall their Ubuntu by formatting the / partition. On my computer, installing a new Ubuntu and the packages I need take approximately 1 hour. Upgrading my whole system takes more than 4 hours ! But, come on, we are not Windows. Do we really need to reinstall every 6 months ?
This one is a lot more important to my opinion. I've installed Ubuntu to a lot of "normal" users. Lemma 1 and Problem 2 or 3 lead them to just think the computer has finished or has a problem. So what will you do if you might have a problem ? Reboot of course, hitting problem 4 in the face !
They will then just discover a black screen with blinking lines of text. Their conclusion is simple enough : the computer is broken and it is because of the upgrade. Technically advanced users not familiar with Ubuntu will try to change X.org config to "repair" the broken X11, never thinking that apt-get dist-upgrade will solve their problem.
That a very bad point for Ubuntu but, more, it has a very important impact : normal users will avoid upgrading at all cost ! The vast majority of normal users I know don't upgrade anymore. They ask me to do it. I'm not speaking about "dist-upgrade", remember lemma 1 ? They just avoid to clic on the orange notification (which is sometimes a red arrow, I don't understand the logic), avoiding even critical upgrade. This week-end, I just came accross a laptop when I was a bit astonished because applications were not available in the repository. I discovered that the laptop was still running Feisty 7.04 because "Upgrade can break things". (for the record, the laptop spend the afternoon upgrading to 7.10 then 8.04 then 8.10). If I had discovered this laptop a few months later, after 7.10 lifetime, I would have no upgrade paths left !
I believe that for 9.10, we will have to think a lot about our upgrade process. Some ideas that should not be that hard to implement :
Also, what would be interesting is to split the upgrade in several "partial upgrade" of a subset of package. On this subset is downloaded, it can be upgraded while the rest is being downloaded. It would also permit to split the upgrade process in multiple days so you are not stuck four hours in front of your computer. But I admit that it might not be an easy thing to do with apt-get...
Not related : If you want to discuss stuffs about Ubuntu or are interested by a "Getting Things Done"-type application for Gnome, just poke me on the GNOME booth at FOSDEM and say "How to get things done in Gnome ?" (I'm more or less like my hackergotchi, it should not be that hard).