Jan 312008

I’ve used gentoo on and off for around 3 years now, since I got my last box, it’s been powering my main production machine at home. I’m a fan of it’s structures and concepts. Sure it’s not really a typical-user distro, but I’m not really a typical user. PClinuxOS is perfect for typical users (and I run it on my laptop where I develop software for them), Ubuntu and many other systems also target that market.
If anything there is a major shortage in hacker oriented distro’s today.

The only two worth mentioning in fact are gentoo and that old standby Slackware. Slackware was my preferred distro for a long time and I know it well, but sadly, it’s age is showing (the details may be a matter for another blog). Gentoo however has rightfully become the distribution of choice among hackers. It’s powerful, fast, and extremely customizable. The source based builds lets you tweak packages to your needs with a powerful and (comparatively) simple set of structures.
While it’s true that portage was inspired by  FreeBSD ports – it is a much more powerful and reliable version of the idea (I know, I use both systems – this server is running FBSD).

And the man most primarily responsible for this is Daniel Robbins – the man who started gentoo. And if you’ve been following his blogs over the past few weeks – it’s scary. While gentoo the distro is strong and powerful – the gentoo foundation appears to be in a downward spiral that isn’t getting any better. First their charter was revoked because apparently nobody bothered to keep their legal papers up to date. Then there was a massive leadership vaccuum – and now there is an election of trustees scheduled. Somebody wanted to nominate Robbins – but apparently he cannot run.
And guess what – users cannot vote – only developers.

Hold on… what is the foundation’s purpose ? To oversee the development and growth of gentoo – surely then to do so with the best interest of the users in mind. After all it’s  a non-profit community project – and if you don’t take care of your users – where will you get your future developers from ?
So the foundation truly must then be said to exist in service of gentoo’s users – yet we get no say in who runs it.
This suddenly makes me a lot less eager to become a gentoo contributor – it makes me scared that gentoo as we know it won’t exist a year from now. This kind of strife needs to be resolved here and now with decisive action by people with genuine passion for gentoo – or it will kill the project. Sure we can always fork it, but rebuilding the infrastructure gentoo had obtained over the years will be an arduous and divisive process – one we should avoid at all costs.

I for one would not mind seeing Robbins back in the saddle at Gentoo  – but whether that is likely (or even possible) I cannot vouch for. Robbins has stated that he will either lead the foundation, or not be involved at all. But to whomsoever gets elected in this election, I hope they will make it a matter of serious priority to resolve all the issues in the foundation fast and effectively. I would hate to see gentoo destroyed – and I hope that whoever becomes the new leaders of the foundation hates that idea as much as me and will do everything in their power to prevent it.

Jan 292008

I run a free software based company. Like any company, that means I have to keep books. I need to keep track of invoices sent and their status, give receipts to customers, keep track of bills paid and other expenses – and be able to see how my company really did financially when all this is said and done.

I tried about 50 different accounting apps over the year of my company, even quickbooks online (shock, shock, horror – I hated trying a non-free program) – the latter quickly failed for me, it is US-useable only. I sure wasn’t going to waste R1K on it’s equally proprietary desktop counterpart – I have my pride after all.
So I started looking for solutions, I even wrote my own invoice creation tool- which worked fine, but didn’t do the rest of it all – and adding it all would be a long, hazardous process.

Then I tried quasar accounting. Finally, I am settled on an accounting app that works. It does everything I want it to do, it’s stable and fast and very feature complete.
Unfortunately the GPL edition hasn’t been updated since 2005. Don’t get too worried, the company is alive and well and so if the project – they have just spent the last two years focusing on adding features to the for-sale parts (the point of sale tool is the major bit). I don’t run a shop, I had no need for a point of sale, so I had no reason to use the proprietary version. The only real impact of this is that to initially install quasar the binary packages are largely useless now – as they are all built for old distro releases. But a rebuild from source worked just fine on pclinuxos 2007 (I haven’t actually gotten it to build on gentoo 64 yet but I’m sure I can).

The biggest grip I have against quasar is one of the most cumbersome installation processes I have ever encountered. The stupid reluctance of desktop distro’s to include enterprise accounting packages in their trees (nothing I checked had it) has a lot to do with this. You have to follow the steps yourself – and there are quite a lot of them.
Make sure you have postgres up and running for starters. The good news is that the steps are very well documented, the installation guide is nearly 60 pages long – but if you follow it, everything should work right out of the box.

Once quasar is installed, the time comes to set up a company. This is complex if you don’t have at least basic accounting skills, but once more there is a very complete manual (the quasar guide) which explains all the various steps and what you are doing there. If, like me, you don’t need all the features (I have no need for inventory for example) then it’s hardly arduous though.
It took me some time to figure out the correct ways to add intransient product items (labour) but by checking the doc and the excellent online help throughout the program, I got through it all.
The process of getting everything set up to match my needs took about 3 hours – honestly from what I hear, thats a very short time compared to what people tell me about programs like pastel. Where I had difficulties it was almost always me not knowing all accounting terminology in English (I only have Grade 10 accounting from high-school and I had it in Afrikaans).
At that point, I could enter my first historic invoice – it took about 10 minutes as it was a first time. Then issue a recept for it’s payment, another 10 minutes to figure out all the details.
From then on, it was easy. Silvia had entered all our historic invoices before the end of yesterday (e.g. in less than a day).

So far, I am extremely happy with the program. It would be nice to see an update to the GPL version, if only to get packages linked to the latest library versions but this is hardly critical. Quasar’s more advanced features are impressive, in fact if you know xml a bit you can completely redesign the interface to your liking.
It’s responsive and fast too.

What I haven’t figured out how to do yet (though I’m sure it’s not hard, I just haven’t gotten there yet) is how to customize the look of invoices to add my own letterheads etc. before I issue new invoices with it, I will obviously need to do this.

The only feature I truly miss from it, isn’t in the retail version either as best I can tell – an integrated payroll system. But payrolling is a very complex thing with enormous amounts of per-country requirements so I also understand why it isn’t there (yet?).
There are many specialized payroll tools out there, so that will be the next step for me, finding one so I can generate accurate payslips before I have to do my tax returns soon. Then I’ll simply enter the actual payments into quasar as expenses paid to the employee salaries account. So despite the lack of a payroll system, with a fairly simple process, my balance sheets will still be exactly right.

Over all I rate quasar a very impressive project. In my book it is the only enterprise level accounting package for GNU/Linux that’s truly up to scratch. Other projects like sql-ledger proved unusably complex. Quasar works with all the enterprise features I want – and it does so in a way simple enough that a non-accountant can still do his books without any major hiccups.

I am now contemplating proposing it as the default accounting package for pclinuxos-BE – it’s much better than sql-ledger as a business owner who uses pclos – it has my seal of approval.

Jan 242008

Well two days into KDE4 I was forced to switch back, at least temporarily, to 3.5.8 – not because I wanted to, but because of a crippling bug that appeared out of nowhere. As of yesterday late, I can no longer add new widgets. If I click "add widgets" nothing happens.

I have not been able to find a source for the bug yet, revdev-rebuild says my system is perfectly consistent.

I also couldn’t print before (oddly, I was able to print from KDE3.5.8) but I have read that there is a problem with KDE4’s print system but it can use that for kde3 so I can investigate this.

Along with the widgets bug though, the system tray stopped working, trying to remove and readd it failed because I cannot add widgets.

I am left with nothing but a big WTF until I figure this out :S

Jan 232008

So I got KDE loaded, and have it working now as per my earlier howto. Having worked with it for a bit now, I’m a bit ambivalent. This is the first ever major KDE release that completely failed to… well inspire me. I can see a lot of potentially – but there is something bland about it. Perhaps a side-effect of the fact that KDE4 had more hype around it than Star Wars the Phantom Menace, but certainly not only that. In all fairness though, that’s a purely emotional response and hardly fair to base the entire review on.

So what then shall we base it on instead ? One approach is a basic ‘pro and con’ list of the new desktop based on my experiences thus far. Keep in mind also that it is built on gentoo so binary distro’s may have resolved one or two of the cons.


First off, despite many complaints from other reviewers, I personally love oxygen, it’s by far the most professional (if a little incomplete) theme KDE has ever shipped with.

The basic black color-scheme appeals to me.

The new compositing kwin is the first compositing window manager that doesn’t cause spectacular crashes on my machine every five minutes nor make it terribly slow.

In fact, this is without a doubt the fastest KDE yet.  QT4 definitely speeded up the general case and things are really much faster.

The SVG effects are gorgeous, the games look absolutely stunning now.

It took me a little while to get used to the new kickoff style menu – but once I did, I really rather like it.


Biggest number 1 con of all: Less than perfect compatibility with my remaining kde3.5 apps ! Firstly there is the fact that everything had to be reconfigured, even when config files existed (I may have been able to bypass this by copying some of the apps directory from kde3.5 to kde4 but that gets ugly).

Secondly, many features no longer seem to work. Konversation’s /media command will not pick up amarok’s current song for example. Amarok lyric plugins all fail. Looks to me like there is a severe backwards compatibility glitch in DCOP.

It is buggy. Not generally severe things, but annoying things, like screen residuals, cursors that dissapear – even the amarok visualization window getting locked inside the middle of my conversation screen and unable to refresh past it for a good ten minutes. The new display code may be powerful, but is is definitely new and still breaks gratitiously.

A lot of people have complained about missing features – honestly they things they list are useful (like adding a second panel) but not critical to me. What did knock me down was that the wallpaper covers the desktop icons, click it right and you can drag the wallpaper away and see it ! But you cannot drag it back – and you cannot see the icons under normal situations. A lot of people complained of missing icons – they are there, they are just underneath the actual wallpaper ! This is a bug that needs to be fixed fast !

So there you have it, a die-hard KDE fan’s feelings on the new release. Overall  like I said, I am not very excited, it fails to grasp the imagination like some prior releases did – but I do like where it’s going. It’s slick and professional and mostly logical – even fun.  I won’t bash and complain that it’s not ready like so many did – I think it was a good time to release, I’ve released things with more bugs in the past (granted, not desktops) and that just helped solve the bugs quicker – this is the FOSS way: release early, release often. I am using it as my default desktop on my home machine now, despite the few annoyances (which I thing speaks for a lot), I will report bugs I find, and do my bit to make 4.1 that shining beacon of human achievement we all want it to be.

Jan 222008

Although [tag]KDE4[/tag] came out last week, I have not yet been able to try it out as I was out of town. Today (Sunday 20 January) however, I decided the time had come, I had some time to spare, and I was a few days behind everybody else. So I started googling…. and what a nightmare. The kde4 overlay that [tag]gentoo[/tag] had, has become obsoleted but the websites haven’t been updated. A thousand sites told me that kde4.0 will not be in the official portage tree – then I finally found out it IS in there (though it took a few days) and it is hard masked.

So confusing was things that at least one blogger gave up and posted a howto that relied on grabbing the source packages yourself and compiling them in a tree, rather than as packages.

Advantage of that method: It keeps it seperate from your existing stuff. Perhaps a lower risk.

Disadvantage: it’s not really the gentoo way is it ? Not to mention it’s a LOT of work.

So having spent ages googling, I felt I would summarize the right way ™ to install KDE4 on gentoo so others can skip digging through litterally hundreds of pages on the forums (not least because the kde4 release sparked the single largest flamewar I have ever seen on the usually amiable gentoo forums).

Finally, please note that I did my tests on gentoo for X64 – YMMV. Now on with the [tag]howto[/tag].

First things first, we’ll need an up-to-date gentoo system:

emerge –sync && emerge portage && emerge –update –deep –newuse world

Short version: Get the latest package list, upgrade portage to the latest version then do a deep world update with your current USE flags. If youhaven’t done a deep world update recently, you can expect some minor issues like blocks to deal with. This is a good time to go make coffee, mow the lawn, make love to somebody and a few other timekillers (it is not fast). If you are brave, you could skip this and run it instead of the emerge kde-meta later on (it will include it by then) – but while this is likely to be faster, it is also likely to break more often.

At this stage you need to get (if you don’t have it) the autounmask tool:

emerge autounmask

Autounmask if you don’t know it, is by far the easiest way to unmask bigger sets of software. With KDE4 hardmasked, you would need to find and unmask dozens of packages, which autounmask will do for you instantly. It’s a very usefull little app to keep for future reference anyway.

Next modify your USE flags a little and ensure that both opengl and xcomposite are in there (by editing /etc/make.conf) – this is required if you want to be able to enable the composition features in kwin later.

You can now test autounmask like so:

autounmask -p kde-base/kde-meta-4.0.0

Check the output, unless you see something very concerning, you can run it for real:

 autounmask kde-base/kde-meta-4.0.0

To get a more minimalistic version, you could use kdebase-meta as well. However, kde-meta will get all the kde4 packages for you. Including the newly redesigned kdegames. There are a lot of packages on that list though, so be prepared to wait a while (again).

You will have to update startkde-3.5.8 first (KDE4 blocks any earlier revisions – the last one is patched to allow it to exist next to KDE4):

   emerge –update /usr/portage/kde-base/kdebase-startkde/kdebase-startkde-3.5.8-r1.ebuild

You also need to force some special use flags for qt4.
Edit /etc/portage/package.use

Add the line:

x11-libs/qt:4 accessibility qt3support dbus gif jpg png ssl zlib


 Finally, it’s time to emerge kde4: 


emerge kde-meta

This will really take a long time. If you skipped the update world above, use this here and now. It is a good idea to do a -pv first, and check if you want to add some temporary USE flags as well. In my case I added:

 USE="pdf svg wmf ppds htmlhandbook debug "

But the list will depend on your current USE setup and needs so decide for yourself.


Once it is finished, you will need to restart kdm to get to a kde4 session:

First edit /etc/conf.d/xdm and change it to read:


Then run:

/etc/init.d/xdm restart

Voila, you can now log into a KDE4 session. You KDE3 session is also still available should you desire it. I’ll post a review of my findings with it in about 24 hours.


Final note: After I started, but before I finished this task  (which took me all of three entire days)- gentoo themselves published an up-to-date howto, I suggest reading it as well for comparison.

Nov 012007

As many of you know, a lot of my business is [tag]development[/tag] using lazarus. [tag]Lazarus[/tag] is critical to my work, which comprises a large amount of object pascal and [tag]delphi[/tag]-like code.

OutKafe is a notable example – and the major source of income for my company.

Which was an issue for me with the desire for my new 64bit box to have a true 64bit OS since I had issues getting it to play nice. On gentoo however, I was able to solve these, and I now have a complete GTK2 (only) 64Bit lazarus development platform setup. This means the upcoming outkafe 5.2.3 due this weekend will be the first to have a native 64bit version( there will still be a native 32bit version as well).

So here are the steps in case anybody else is stuck.

1) Check out the latest lazarus from SVN. Save that in a folder somewhere

2) Emerge FPC with teh SOURCE use set:

USE="source" emerge fpc

3) Make sure you have GTK2 already installed (emerge gtk2) as we will have to use nodeps in the next bit.

4) The current lazarus ebuild wants to downgrade you FPC (I think that’s a bug) so do this: emerge –nodeps lazarus

5) If you have an old .lazarus directory in your $HOME – wipe it.

6) Copy the svn lazarus you checked out OVER the newly installed version in /usr/share/lazarus

7) Run lazarus with: ‘startlazarus’

8) Rebuild lazarus with the internal menu (only for GTK2) to build the updated source code.


Voila, you are done.

Sep 152007

Yep, you saw that right. PCLinuxOS has bumped Ubuntu from the top spot on distrowatch. This is the first time since the release of Hoary Hedgehog several years ago that [tag]Ubuntu[/tag] has not held the number one spot. This blogger seems extraodinarily surpized by it, citing all the things which Ubuntu does right and [tag]PCLinuxOS[/tag] does wrong. But he doesn’t mention what PCLinuxOS has done right all along, and Ubuntu never did. See there were basically three generations of Linux distributions in history. The very first generation was largely confined to programmers and consisted of basically building the whole thing from nothing. Technically these were not even really distro’s. Modern source distributions like Gentoo cannot be considdered first-gen btw. since they do have structured install methods and packages. The second generation started with Slackware really, and the main difference was that they used prebuilt packages to install the distribution’s various software pieces. Distro’s like redhat, suse and Debian are still 2nd generation today, and 2nd-gen distro’s are the best server distro’s around. However, since the turn of the century a third generation began to appear.

These are primarily recognizable by one thing: they ship as liveCD’s, which can be modified live while booted, and installed ready to use (with such modifications in place). This has massive advantages for a desktop distribution – particularly as it lets the user verify his hardware compatibility prior to the actual installation. The major groundwork that would lead to 3rd-gen distro’s was laid by knoppix, though knoppix never really considered installation important and therefore never really became a complete 3rd-generation. It could be installed, but this was difficult and even today is largely manual.

To the best of my knowledge, the first true third-generation distro was OpenLab 4.0 – which I designed (due credit, I had a lot of outside influences). This made OpenLab actually a successful contender against the might of Ubuntu’s marketing forces. But OpenLab was just the first, since then a number of other distributions have followed suit. The best known being PCLinuxOS and SimplyMepis. These were true 3rd-generation distros. PCLinuxOS even shares with OpenLab the design concept of having a login screen on the LiveCD where a guest user is just a normal livecd experience, but the root account is used for installations.

Third-generation distributions do not leave the concept of packages behind, they are still used for the user to install add-on software later, but skips the complexity of packages for the basic install, providing a solid set of pre-installed and preconfigured packages right from the get-go.

Ubuntu is still not a true 3rd-generation distribution, although it did copy the basic feature in some later versions this remains unstable and Ubuntu states that it is still not the prefered method of installation.

It doesn’t end there. Ubuntu has a habit of ignoring bugs that aren’t fun, leaving users to struggle with things that (to them) are critical. For the longest time Ubuntu users had no way of graphically setting up a pppoe connection, in fact to do it at all you had to download a package… hold on – that’s right, to get online, you had to be online first (by comparison, OpenLab had this included before Hoary Hedgehog was ever released)

This lack of attention to detail is why Ubuntu never deserved to be in the topspot. Ubuntu was always a lot of hype, never anything else. I couldn’t say this when I was a distro developer, but now, not being one anymore I can give an unbiassed view that Ubuntu was never all that good and in fact the number of newbies who were referred to it probably did Linux growth more harm in recent years than even the Novell/Microsoft deal. I stand by my belief that there is no better desktop distribution available today than PClinuxOS.

Having said that, I don’t believe Ubuntu is always a bad distro, I myself recommended kubuntu to a customer on a major recent project. But, I knew I would be the one setting it up. That changed everything – I knew that no linux newbie would ever have to deal with the first six weeks of life with Ubuntu. Everywhere else, I will keep recommending PCLinuxOS until there is something truly better (which may be just the next version of it).

It took a long time, but finally, people are starting to see through the hype, and PCLinuxOS is holding the postion it should have had all along.

Jan 102007

It’s simply a fact that one needs to do comparisons with other distro’s in my field, frequently.

Since I have two personal machines in daily use, it is easy to try out distro’s on one while maintaing a bleeding-edge OpenLab on the other. Currently I haven’t really made up my mind on what I want to try however. One side of me is tempted to try gentoo 2006.1, another side to try two desktop favorites: kubuntu and pclinuxos.

Especially the last one is drawing me, since I have heard people say “OpenLab and pclinuxos are currently by far the two easiest desktop distributions out there” – if these guys are our match, then I need to see what I can see.

So what shall I do ? I don’t know, I do love the gentoo concept – but it takes a LONG time to do a full install (at least a level 3 one and what’s the point otherwize ?) and it appeals to my geek side- not an easy sell though.
Desktop wise, well it’s between kubunto and pclos…(no I am not and never will be interested in trying gnome as my personal desktop again, I don’t like the basic premise on which it is designed – dumbing down != easier to use) …
I reckon I’ll give both a shot, and keep the one I like best… for a while at least.

Update: I went with PCLinuxOS – and I am very impressed. This distro is shuweet. Kudos from the silentcoder.