hardware database and testing

This is going to be a short post about open shared hardware databases.

Since the early days of open GNU/Linux open development and practices i.e. since the 90’s it was painful for a new user to know before-hand if some system would run their favorite GNU/Linux distribution or not.

By running I mean :-
a. would it show X windows (the GUI) or would it throw on a command-line mode?
b. does the mouse respond/work ?
c. Even if runs graphical display does it do at the right resolution ?
d. Does the display take the whole space offered by the monitor ?
e. Does the hard disk perform adequately as far as speed, performance are concerned ?

and whole lot of questions.

I have had talked about this hardware experience/inexperiencearound 2 years back but in a more stub-like manner (very short) .

The problem for GNU/Linux distributions is they, the developers and maintainers of the distributions do not know what hardware are their users using and running and which hardware fails or has issues for most of the users, the majority. The ones which affect more would be most urgently hacked upon by developers by reverse-engineering the device and how it talks, or/and talking with hardware manufacturers to give them info about it, hardware specifications and what not.

For instance, one of the more common or uncommon use-cases/usage which we see as GNU/Linux failing is when people have to resort to using ndiswrapper to do their hardware. It would be definitely interesting to see/know if number of devices have grown or shrunk which need to use ndiswrapper.

Unfortunately, the situation will has improved vastly over the years they still happen from time to time as illustrated quite recently

There have been many different approaches to tackle this issue. Right from having system manufacturers making libre machines like System76 which give the option of having Ubuntu (a popular Debian-derived distribution) pre-installed on desktops and laptops. There is also Dell which sells pre-installed Ubuntu laptops. The problem with the former is they are only targeting the States and Canada (which leads the rest of the world dry) while the problem with the latter is that there is not much of a price difference as compared to the Windows based-machine. Also the support for dell for Ubuntu leaves much to be desired.

While getting some computers having pre-installed GNU/Linux is a good move, it’s just not enough. Even in Dell you would see they have a low-range offering and no depth as compared to their Windows offerings, both in the laptop as well as the desktop space so a prospective user would end up making a compromise on his choice.

Another point is that While Ubuntu has a lot of good things going for it, there are lots of other good distributions like Redhat‘s Fedora, Opensuse, Debian GNU/Linux distribution and its ilk which do not figure in any of these companies plans which means a lower uptake.

The number of people who would install any OS would be very small rather than people who would love a pre-installed experience. No more hunting for drivers, everything works out of the box (at least that is the dream the manufacturers sell) and one can get on with doing what they want to do in the first place. This is a piece of puzzle that the dealers and manufacturers of hardware would need to solve in order to have a choice and also leverage the behemoth called Windows. It also would question and bring some much-needed diversity to existing paradigms of Graphical User Interfaces and technologies.

As of right now, its MS-Windows which calls the shots on an uptake of any technology. It’s a sad fact but it is. This delays new technology reaching to the masses and unsurprisingly the cost of the computing remains high.

Another part of the puzzle (notably graphics) lies with both distributions, companies and hardware manufacturers talking in and about common interfaces. See for instance the recent move to Gallium3D from everybody (else) except Intel, surprise surprise.

Another interesting part of the puzzle which the distributions are trying to solve is with the help of making hardware databases. Having a list of hardware which most of their users use and trying to get an idea which of this hardware fails or is inefficient. There are four different projects I know of which are trying to deal with this issue.

a. smolt is a fedora project which tries to get a sense of what is the base hardware that most of the users are using and how the trends are changing.

b.HWDB is Ubuntu’s attempt to solving the same issue.

c.hardware4linux is another attempt to do the same thing. unfortunately no native debian packages till date.

d.h-node is the newest kid on the block.

Unfortunately all of the above are eccentric and do not talk to each nicely. While smolts has fedora users and opensuse users putting their hardware profiles in the smolts database, ubuntu choose to go with hwdb .

Also both smolts and hwdb do not disclose all the data/transparency some for marketing purposes and fear. While in the desktop space both may have similar market shares, it is in the workstation and server markets (which has traditionally been a GNU/Linux stronghold) they probably do not want to share the data. This whole issue can be more properly shared if one reads the smolts request for package bug report in Debian.

The same issue can be seen with Ubuntu’s hwdb database. For a potential user, till date there is no good pre-existing hardware guide which would tell which hardware is good at each different budget point so people can make decisions easily. Each upload to the database is hashed and potential users cannot view what hardware is good/bad uneven so they are blind. Also if people think Live CD‘s do solve the problem, it does not. I have come across various instances where a graphic card, a sound system worked/works perfectly in the Live CD mode but after installing one had to do a lot of fiddling around just to make it work.

While I did share about hardware4linux.info one of the things apart from the fact mentioned above is they do not many profiles, something like 4,500 profiles till date.

The youngest kid on the block is obviously h-node and it also suffers from a similar issue. A more fundamental ideological issue. The freeness of distributions. It will let only those distributions upload their hardware stats who fit under GNU’s free system guidelines. Unfortunately, Debian does not figure in that list making it another yet not mainstream movement.

The upshot of this was that after a long time I took a look at what components are in this system which are from non-free.

For instance, I knew that Opera would certainly fall in it (I use it to test/see some sites when they do not render properly in either Mozilla Firefox and Google Chromium) . I was under the assumption it would be the only one.

Much to my shock and surprise there were quite a few more :-

$ aptitude search ~i~snon-free -F”%p# %s# %t#”

firmware-realtek non-free/kernel testing,unstable
gcc-doc-base non-free/doc stable,testing,uns
icc-profiles non-free/graphics testing,unstable
opera non-free/web stable
sun-java6-bin non-free/java testing,unstable
sun-java6-jre non-free/java testing,unstable
unrar non-free/utils testing,unstable

The contrib section has only package installed on my system

$ aptitude search ~i~scontrib -F”%p# %s# %t#”
ttf-mscorefonts-installer contrib/fonts stable,testing,unstable

From the above, while firmware-realtek, icc-profiles and unrar are correct but slipped out of my mind, I am still shocked to find gcc-doc-base and sun-java6-* to be still in the non-free section. I was thinking/was of opinion that Sun Java had become free. Did Oracle make it again non-free ?

I hope to remove mscorefonts and replace it with the ubuntu-font-family but it has to go a long long way to go. I have done a local install of ubuntu-font-family.

$ aptitude show ttf-ubuntu-font-family
Package: ttf-ubuntu-font-family
State: installed
Automatically installed: no
Version: 0.71.2-0ubuntu5
Priority: optional
Section: fonts
Maintainer: Ubuntu Developers
Uncompressed Size: 3,121 k
Depends: defoma
Conflicts: ubuntu-private-fonts, ubuntu-private-nda-fonts
Replaces: ubuntu-private-fonts, ubuntu-private-nda-fonts
Provides: ubuntu-private-fonts, ubuntu-private-nda-fonts
Description: Ubuntu Font Family, sans-serif typeface hinted for clarity.The Ubuntu Font Family are a set of matching new libre/open fonts in development during 2010–2011. The development is being funded by Canonical Ltd on behalf the wider Free Software community and the Ubuntu project. The technical font design work and implementation is being undertaken by Dalton Maag.

Both the final font truetype/OpenType files and the design files used to produce the font family are distributed under an open licence and you are expressly encouraged to experiment, modify, share and improve.
Homepage: http://font.ubuntu.com/

Anyways, that’s about it for now. As seen today, there is still quite a bit of struggle that all the distributions need to sit across and come to terms with. Till there is no united front, we cannot hope to compete with MS-Windows in our islands. Wonder what the Linux Foundation and Free Desktop people doing about this, this should be their baby as both want more GNU/Linux adoption.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.