RedHat 9

OK, bear with me a mo' - I'm just rebuilding gizmo to be a RedHat 9 Linux box, and the upgrade is *way* less smooth than a clean install. I would thoroughly recommend against doing an upgrade from the CD with RedHat I usually get better results by copying everything to a second disk on the same PC, then installing from scratch but leaving the backup alone. Then you can copy everything across (like /etc/X11, etc) as and when you need it.

Basically, if you have a system which has elements which aren't in the RPM db, you'll need to take a lot of control over the process.

I just downloaded RedHat 9 over the long Easter Weekend just passed and I've spent the week trying it out on my work workstation. I haven't used Windows very much in the last few days, and didn't use it at all today, except to copy some config files around. I use Evolution linking to our Exchange server and that's amazing. I remember seeing early versions of that which were scary, but right now it's what Outlook should be. It has all the basic functionality of Outlook (connection to and Exchange server and LDAP directory stuff) and the adds a ton of features (turning off image loading, threading of messages.) I really have trouble believing that the most common email client can't thread a message. Sure, you can group by conversation, but that's a useless feature in practice. Threading is what you need for mailing lists, any involved email exchange, or just keeping track of who said what to whom and in response to what.

There were three critical apps keeping me using windows at work. Outlook (replaced by Evolution as above), Agent (replaced by Pan, although I downloaded a more up to date rpm from the pan site), and Excel.

The most recent versions of Pan have most of the usability of Agent and they have score files. They still don't have the 'next message on middle mouse button' features, but, score files more than make up for that. I've been using Agent since 1996 or so, but the development of it is too slow, and the developers are catering for a binary-downloading user (i.e. porn hound) rather than someone who wants to participate in one of the few communities left on Usenet.

Excel. Nope, open office still doesn't touch it. Nothing touches it. Excel is Microsoft's killer app, and they probably don't even know it. It hasn't changed very much for about five years, 'cause there's not much to change. I've sent at least two large firms down the Microsoft Office route since I started using it in 1993 or so. Just today, being determined to have a Microsoft free day, I wanted to chuck some numbers I grabbed from the screen into a spreadsheet, parse them into individual entries and see how they related to each other. Now in Excel, alt-d, e brings up the 'parse selected cells' dialog. Open Office calc didn't seem to have that. I couldn't find anyway to split a cell by spaces (or whatever) without resorting to writing a BASIC program. And that's an enormous shot in the foot right there. Sure, you're copying Excel by having BASIC as the macro language, but, often, that's the only programming language available on a Windows box. And VBA (Visual Basic for Applications) in Excel is a great tool for reading in data, processing data, writing out results files - it's the reason why Lotus 1-2-3 no longer exists and why Quattro never took off - don't get me wrong. But this is a Linux box. It has Perl, Python, C, C++, Fortran, Bash, etc all available as scripting/programming languages. Let me use Perl on a spreadsheet and Microsoft will wobble. Let me use a combination of C, Fortran and Perl on a spreadsheet and I can eliminate proprietary transport planning software almost straight away. All it would need would be for the spreadsheet to recognise the #! declaration and insert a library of it's own to link in. e.g.

 #!/usr/bin/perl use OpenOffice;  ... my $row=10; my $col=100; foreach my $parameter (sort keys %value) {  print SHEET("1999 Results",$row,$col) "$parameter";  print SHEET("1999 Results",$row,$col+1) "$value{$parameter}";  $row++; }  

If you know perl and associative arrays, you know just how powerful that would be. If you don't know perl, it's hard to describe, but there just isn't a better programming language for analysing things where the data structures have to be built on the fly.