IBM's Interest in Red Flag Linux

eldavojohn writes "For those of you unfamiliar with Red Flag Linux, it's an OS for the growing Chinese community of Linux users. Interestingly enough, IBM is looking to support Red Flag Linux as the next distribution of Linux that its more than 300 applications will run on. Support from a huge vendor like IBM certainly raises the rate of adoption of a distribution of Linux so this is certainly good news for Red Flag Linux and also the Chinese open source users. IBM currently supports Red Hat and SUSE Linux, which creates twice as much testing for each of their applications. Will Red Flag Linux cause them to require three times the amount of normal testing?"

This is a good thing.

[miffo.swe]

If IBM is smart they will target LSB (Linux Standard Base). Then they will ask the distributions to please conform to that standard. If anything this is the kind of thing that could work on unify Linux even better if done right. Ofcourse testing will have to be done anyway but the likelyhood of problems will be very small for every new distribution supported.

Asianux

[raffe]

Red flag is based of Asianux which is based on red hat.The current release version of Asianux is 2.0 , which is based upon Red Hat Enterprise Linux 4. Asianux 1.0 was based upon Red Hat Enterprise Linux 3. The first releases of red flag was very poor quality. Asianux is a joint development between Linux vendors Red Flag , Miracle Linux Corporation and Haansoft .

Re:Red Flag...LSB compliant

[ronanbear]

A list of Linux Standard Base compliant distros is available on their site http://www.freestandards.org/en/Products [freestandards.org] Asianux 2.0 (which Red Flag is based on) is listed as compliant so it would be reasonable to assume compliance. I've never used Red Flag so YMMV.

Re:This is a good thing.

[teg]

RHEL and SLES are both LSB certified already - the problem is that LSB doesn't specify enough to be useful. An LSB-compliant application needs to include everything outside the LSB scope in itself, which ends up being the OS minus the X libraries and glibc (I'm exagerrating, but not that much). LSB is the lowest common denominator, and a very static target, in a world of rapid evolvement (e.g. GNOME every 6 months, new compilers, new glibcs etc).

Re:times as much testing?

[anzev]

This is not true. Testing is time consuming because in a commercial environemnt you HAVE TO test ALL THE FEATUREs on ALL THE SUPPORTED PLATFORMS. Testing is not what most users think, that you just run the app, click a few buttons and that's it. The common mentality is, as it appears:

"Ok, let's test this on OS A, and if there are errors fix them, otherwise, see if the program also runs on other OSes and that's it".

Sorry to dissapoint you, wrong! You need a test plan [wikipedia.org]. The test plan specifies how to test each feature (steps to acomplish it). These test plans have to be carried out on any given platform that you mark as SUPPORTED. There is the obvious difference between supported and "it also runs on". Maybe a quick example of why testing the features is important. Java for example, on Windows and on Linux behaves differently. In particular, it treats String objects differently. On Windows it is legal to compare two strings like this:

if(str1 == str2)

But this approach fails on Linux. And it's also technically more correct to write:

if(str1 == null || str2 == null) throw new IllegalArgumentException();
if(str1.equals(str2))...

Even testing between different distrbutions can be a problem due to versions and "uncontrolled" API of open source applications. In our company we had to test (this is execute the test plan) on a few different Linux distributions because we tested it on Debian and it didn't work on Red Hat!

Re:IBM ?

[leoPetr]

No, the article means IBM, which develops the DB2, Informix, Rational, WebSphere, and Lotus ranges of software products. Lenovo is the new manufacturer of Thinkpads. They are not the new IBM.