for some time a lucky few inside sun have been playing with AMD64 (athalon) acer ferrari 3400 laptops . we've been enjoying quick-turnaround on added hardware-specific features like the battery monitor, power management and more thanks to a casper and his frkit (ferrari-kit) utility. now you can play with it too thanks to opensolaris and the CDDL.

there are lots of reasons you might want to change the default Solaris 10 login screen. here's the quick and easy instructions for doing it.

the file which contains everything you want to know is called 'styleModern'. making changes to this will create the effect you want without mangling too much of the standard config.

first, create direct /etc/dt/config/C, and copy in styleModern:
# mkdir /etc/dt/config/C
# cp /usr/dt/config/C/styleModern /etc/dt/config/C/styleModern

next, create a local images directory to hold your modified images, and copy across some of the basic items (which you won't customize)
# mkdir /etc/dt/config/images
# cp /usr/dt/config/images/A* /usr/dt/config/images/S* /etc/dt/config/images


now create your custom dtlogin backdrops as gzipped xpm files. you need one file for each of the likely screen sizes, and each file must be named in the format name-size.pm.gz. for example, mybackdrop-1400.pm.gz for 1400x1050 (laptop) screen, or mybackdrop-800.pm.gz for an 800x600 screen. save these files in /etc/dt/config/images .

# ls /etc/dt/config/images
Adobe.im1.Z mybackdrop-800.pm.gz Solaris.im8g.Z
Adobe.im8 Solaris.im1.Z
mybackdrop-1400.pm.gz Solaris.im8

now edit your /etc/dt/config/C/styleModern file, and make a couple of changes. change the line defining PIXMAPDIR from
#define PIXMAPDIR /usr/dt/config/images
to
#define PIXMAPDIR /etc/dt/config/images


and change the line defining STYLE_PIXMAP from
#define STYLE_PIXMAP PIXMAPDIR/styleModern-backdrop
to
#define STYLE_PIXMAP PIXMAPDIR/mybackdrop

(or whatever you set 'name' to in your files above.

save, logout, and enjoy the new backdrop!

to revert to the default, simply delete the /etc/dt/config/C/styleModern file

last night's inaugral SOSUG meeting seemed to go quite well, with about 20 people enjoying presentations from brendan gregg and alan hargreaves on dtrace and SDT probes respectively. brendan's was probably the slowest intro to dtrace i've ever heard ;) - jon haslam and bryan cantrill are hyped on on something (excitement?) whenever i hear them talk and seem to want to fit a 4-hour presentation into 20 minutes 8). (they're all great, though, and fun to watch to see if - for example - jon will actually explode on stage). brendan on the other hand actually seemed to talk like a normal person, although he couldn't help getting excited either as he demonstrated just a few micron's of this tool's abilities.

the next SOSUG meeting will probably be in a month's time. if you're going to be in sydney around that time - check out the sosug google group to see what the details would be. there was a nice-sounding idea about holding it at the james squire brew house, which apparantly has meeting rooms with wireless hotspots.

mmmmmm. beer.

many moons ago the first degree i started at university was architecture (you know, buildings). at the time "IT" professionals were in EDP or MIS departments, and spend many dull days fiddling about with COBOL code: not something i really wanted to do - for me computing was a hobby. as it turned out, architecture is one of those things you either need to be amazingly talented at (i wasn't, particularly) or have a great deal of life experience for (i certainly didn't) - otherwise you spend your working life spending many dull days fiddling about with window details or material schedules. so i quit & fell into IT as a systems operator.

however, all was not lost from those 3 years of studying architecture - it helped emphasise the importance of effective communication, including visual communication and the subconscious communication of interaction with the environment. this easily transfers into the computing world with the simple question:

is is easy to use?

the problem so far, unfortunately, is that there are many and varied definitions of the extremely subjective term "easy". trying to define it in a computing context to make everyone happy will almost certainly start a flamewar, no matter where the conversation is held. here's the start of my definition though:

are simple things simple to do?

in practice this means you should spend (considerably) less time setting something up than you spend enjoying the result; and if a someone understands the procedure for achieving result 'A', they should be able to understand or at least guess the procedure for achieving similar or related result 'B'.

what has this got to do with solaris? well - it would be disingenous to claim that solaris has targetted "regular" users up until recently. solaris 10 has made huge inroads into this however, with a modern GUI, built-in office productivity suite, and many other tools. it's also made huge leaps in performance and observability, which means there's an opportunity to have a major focus on how people approach using it.

i have my own ideas on what's important to look at, but i'd like to hear others. leave comments.

today i installed the latest internal build of solaris onto my laptop and it is nice.

this build has integrated the new boot methods which casper has mentioned before. in brief, this means that the booting and installing solaris is a much quicker and easier experience on x86/x64 hardware than it has been in the past, and should open up the door to booting from all sorts of interesting devices (people are talking about booting from USB flashdrives and so on).

the cute thing about the new solaris boot is that if you don't like the text console messages popping up you can do away with them altogether, replaced by a rather spiffy solaris splashscreen. all this and more should appear in the next solaris software express release (the one after 4/05).

and of course, opensolaris™

i don't usually read slashdot and almost never read the comments to the main articles, and today i reminded myself why. this thread contains so many wild and ludicrous claims of how sun operates and what is and is not inside solaris that i felt my blood pressure shooting through the roof after only 40 seconds of reading.

most of the ludicrous claims revolve around the assertion that opensolaris™ is vaporware. well, sun's spent a lot of money, effort and corporate reputation on getting this going, so there's no way it's not going to happen. if you're worried about timing then there's a simple answer: other people's IP. the code has to be so clean you could eat your dinner off it. otherwise something like this or or this could happen. ever wonder what would have happened if IBM had sued SCO over a similar issue? (here's a hint - IBM has more money & probably a bigger legal team...) now consider how much some of sun's competitors would like to see sun microsystems "disappear".

so opensolaris will be ready when it's ready - so sun can keep on trading without lawsuits popping up to distract.

but back to other ludicrous claims in slashdot. we had:

• " Maybe you don't see much difference on your SS5, but on my 15k domains SVM/UFS can't hold a candle to VxVM/VxFS." - my response is "something is probably wrong with your UFS set up then" (see here)
• "Sun is probably going to try to shoehorn the Linux JFS/Ext3 filesystem into the kernel. I guess they could just release it with UFS only, but who'd want to run a server on a non-journaled filesystem?" - at which point i almost fell off my chair. the phrase rtfm came immediately to mind. ufs logging has been around since 1998, but why let facts get in the way...
• i got bored looking now... but maybe you get the point

i wonder if the ranters are ranting because they can't wait - or because they just like ranting?

... no seriously...
there's often a lot of talk about how linux (in and of itself) is such an innovative OS and how it's development is so much faster than other platforms.

well duh. when you're cloning or reimplementing something that's already been done it's a much easier task. making something up from scratch - now that's harder.

so i'd really like to know - what has linux actually provided to OS technology?

send comments, then i'll shut up.

a couple of weeks ago simon phipps flew into sydney for a meeting with some folk looking to start an opensolaris^™ user group in sydney. the meeting has borne some fruit, and a sosug google group has been set up. so - if you're in the sydney area and want to be involved in such a group, go sign up. now. we'll still be here when you get back.