Older blog entries for Burgundavia (starting at number 112)

Rocking towards a more unified GNOME

A few of us were talking on #gnome-hackers on Freenode about vendor patches. Davyd, the gnome-applets maintainer, was cursing various vendor patches. As we started talking we realized that there is no single place for GNOME maintainers to go looking for vendor patches. After Glynn Foster of SUN offered up JDSs unified patches and spec files list (direct link to the patches against 2.14), we realized we needed such a central place. Thus VendorPatches as born. As can be seen, OpenSolaris, Ubuntu, Gentoo and OpenSUSE and DiscoverStation are already listed.

The next iteration of that page would be present direct links to the actual patches themselves, which involves parsing a bunch of arcane source formats and presenting them in a nice neat view. All in all, less vendor patches can only be a good thing.

As an aside, isn't this what Launchpad was supposed to solve?

Don't compromise my freedom for a spiny cube

Apparently Feisty is going to ship binary drivers by default. That burning effect everybody was apparently commenting about in Beryl? That was a brilliant opportunity being burned away. The opportunity to win truly free drivers. ATI/AMD are only going to release free drivers if it is in their economic best interest. If the most high profile distro decides that binary is "free enough", this seriously hurts anybody advocating it being open source, both in and outside ATI/AMD.

Why listen to me? I work for a company that has created a great deal of closed source stuff on top of Linux, including a fairly nice multiseat X server. Currently that X server is in multiverse (package is desktop-multiplier), rightly so. I hammer away at our managment, explaining that we won't get anywhere in the Linux community without it being open source. What if Ubuntu decided to use our server instead of upstream X? I would look like a complete fool to company management, because I was clearly misleading them. Efforts to get it open sourced would be seriously compromised by my loss of credibility. I suspect that is exactly what is going to happen with Feisty with the ATI/Nvidia drivers.

Please, lets reconsider this. Lets make a public statement, clearly telling ATI/AMD and Nvidia that we value Freedom over any shiny cubes or fancy bling. Lets support those brave people within ATI/AMD that are helping get us specs, not shoot them in the back right on the finish line.

10 Nov 2006 (updated 10 Nov 2006 at 06:36 UTC) »
On the consistency of tabs

Tabs in most apps are not very consistent. Different shortcuts, closing buttons, ways of programmatically making them, etc. Making them consistent is good. As such, the Tab Consistency spec as an idea is a brillant thing. However, there is a small problem with the spec as currently written: It simulteanously changes the way every tabbed app on Ubuntu works (including Firefox 2) back to the old Firefox 1.5 way , introduces some pretty heavy patches to do it and ignores a bunch of usability work that has been done in this area.

The bad

  • The "new tab" icon should itself be on a tab (which can be tiny - just containing the "new tab" icon) which is on the left of all the tabs. The "new tab" icon should use the same symbology between desktop platforms. Clicking several times in succession on this "new tab" tab would create the corresponding number of new, empty tabs. The rationale for having the "new tab" tab on the LEFT is that it keeps it far away from the "close tab" icon (which will be on the RIGHT of all the tabs), and keeps it in a consistent place. The other option was to have the "new tab" tab on the RIGHT of the current set of tabs, but this means it will constantly move around, meaning there is no opportunity to build up muscle memory for the "new tab" click.

    Comment: Having a button to open new tab is sane. Having it in the tab bar probably is not. No currently tabbed app in any OS (that I know of) does this nor have I seen anybody even propose it. Interesting idea, but it would essentially mean doing usability testing on 6 million+ Ubuntu desktops with no feedback. Not my idea of fun.

    Update: I was pointed at Konsole, which has such a button. Still undecided.

    2nd Update: Jeff Waugh pointed out that IE7 has such a tab as well.

  • There should be a "close tab" icon on the far RIGHT side of the tab bar (the bar on which the tabs appear). Comment: This is the old Firefox 1.5 model, which Mozilla explicitly moved away from in 2.0, due to Google-funded usability studies. You are read about them here. Why is this bad? It violates the concept of a tab as a physical object (which I will illustrate better in the moving tabs part. Yes, it does lead to accidental closures. However, from my experience and that of the Mozilla usability tests, I think the confusion over "how to close a tab" is a bigger issue than the "I closed a tab accidentally". Note that Opera and now Firefox get around this by allowing trivial reopening of tabs. Not saving users data is a bug. This is not a fix for that bug

  • "It should be possible to move a tab around in the sequence of tabs, by dragging it with the mouse. A small arrow should show the space between two tabs into which the current dragged tab would be placed. The arrow should be themable and consistent across all tabbed applications." Comment: This is currently how Firefox does moving and it has always irked me. I realized today why. It violates the whole "Tabs as objects" principle. When you move a physical object or a window, the object or window just moves there. It does not draw a rectangle or an arrow where I want to move it, it just does it. I can illustrate it very neatly with the following gif: . As you can see, when you move the beer bottle, it just moves. (Sorry, I had to get beer in there somehow). Having the tab just move has always "felt" more natural to me and the key part: It is faster and more fluid. Why? For one, the little arrow is tiny and it jumps around, rather than moving naturally. This means it is hard for your eye to follow. A full tab is quite large and moves cleanly.

    Update: I should clearly state I am talking about the Epiphany way of moving tabs. gnome-terminal just jumps around with a clear indication of which way it went.

There there is the issue of upstream. There are two major ones that care here: Mozilla and GNOME.

Mozilla - Given we just signed a nice agreement with them about working together and one of their stated goals is a consistent look, what are they going to say?

GNOME - The list of GNOME apps include Epiphany, Gedit and gnome-terminal. While tab consistency is very much on the agenda, especially given GTK finally got notebook support, I don't think this is what they had in mind. As the spec freely admits, if we implement this spec we are likely going to have to carry patches around forever and may break API stability (a key GNOME promise).

The good

This spec is not all bad. There are some good points about consistency in the GNOME apps, seen under the Implementation section. There is also a good discussion about keyboard shortcuts. Both of these parts should be talked about and implemented.

In closing

Sebastian Bacher is probably going to ask me why I am raising this now. Very simple. The spec is, as of this writing, at "Pending Approval". Given that this is Mark's spec, I wouldn't bet against it being approved. That means, time willing, this will be implemented for Feisty. As this is the spec I completely disagree with the implementation one, I thought I would raise my specific issues.

In the end, tab consistency is a good thing. However, the proper place for all this discussion is upstream, not in the distro. Let start a dialog in GNOME, decide on a course and all work towards that. Vendor patches are not going to make anybody happy here.

Tomorrow: Rocking specs from Mountain View (there were tonnes and love most all of them (except this one)

9 Nov 2006 (updated 10 Nov 2006 at 02:05 UTC) »
Please don't fan the flames

Steve, please don't fan the flames. The update to DaveJ's blog came because I emailed him politely and pointed out his error.

As for the OLPC stuff, it has been made abundantly clear that the shipping OLPC image will contain completely free software, including no binary blobs from Marvell. The binary blobs are there right now so the boards can be tested while the free firmware is written. I can dig up a reference for this if you want. Remember that the wireless card is completely new, not standard consumer grade stuff.

For the record, I do think we could do better getting stuff upstream. More on this (and Useful's challenges in this area) later.

"We are pleased to inform you that you are now part of the GNOME Foundation Membership."

Rock! I was a little worried my membership application had been lost in the shuffle but I guess my concerns were for naught. Now where is the membership committee so I can join and make certain other people don't wait was as long as I did?

8 Nov 2006 (updated 8 Nov 2006 at 00:08 UTC) »
Integrating GNOME into Active Directory

Integrating a Linux desktop into Active Directory is one of the dreams of nearly any Linux admin. The simple fact of the matter is that AD is a very good product and has won in the marketplace. Just hooking up your Linux clients and having them authenticate against AD is not hard, and the ubuntu-directory team is making it easier. However, this is only part of the puzzle. As Jorge Castro said to me, "I would kill to be to control the settings on my Ubuntu clients from AD".

Jorge, start your murdering. Today, Enabled People, makers of Linux-XP, showcased their "Integration Service", which purports to show the controlling of gconf lockdown via AD policy. Interesting, but no code yet, let alone a license the code will be under. We can only dream...

7 Nov 2006 (updated 7 Nov 2006 at 20:06 UTC) »
Are you voting?

If you live the US, you need to vote. My grandfather used to say to me "You can only complain about the government if you voted". And the only way we are ever going to get a sane US government for the sane to go and vote. Don't know how? Your Guide to Voting

If you possibly can, vote via paper ballot. Electronic voting machines have serious issues and don't buy the argument that it will take weeks to count your vote. Canada has managed to vote via paper ballot for quite some time now and still delivers accurate results on the night of the election.

1 Nov 2006 (updated 1 Nov 2006 at 03:36 UTC) »
Being nice while creating teams on Launchpad

Please, when you got and create a team on Launchpad, please append the ubuntu- prefix to your team. Thus if you are working on product foobar in Ubuntu, your team is not foobar but ubuntu-foobar.

Why, you ask? Two reasons, namespace and findability

Namespace: Because Launchpad is used by more than just Ubuntu, you don't want to run into a conflict. What if the original authors of foobar want to use LP?

Findability: If you team name contains ubuntu- in its name, it is going to be abundantly clear that you mean foobar in Ubuntu, not just foobar in general.

Yes, long time, no blog

Life happens. A "What I did this summer^WWWfall" coming soon.

7 Sep 2006 (updated 7 Sep 2006 at 00:38 UTC) »
Chiming 16 past 2

Ubuntu everywhere

Even on Novell's developer pages: How to configure Ubuntu for AD. Thanks Ryan for the link.

You can't fault the Novell guys for trying and SLED 10 isn't a bad product. It is just that you just can't buy a community like Ubuntu's, nor can just expect one to grow absent any structure. This is something which has become more and more apparent in recent weeks. You need things for it to grow on and gardeners to channel the growth into some sort of order. But mostly you need the right frame of mind. As Jane Silber said at Ubucon, "We started by getting the community right and then the company, rather than other way around". It says a lot that the COO of a company would come to a small community gathering and not be there to tell us about how great the company was (well, maybe we already knew that...). In fact, the only time the Canonical people took the floor was when asked to and then only for a few minutes.

103 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!