James Bottomley speaks

Dealing with high-powered nerds all day must have yielded some insight into how to deal with big brains and big egos. Any social suggestions for people who have to make groups of geeks function together?
Well, I’ve got several lectures on this one. In Linux (especially the kernel), being a maintainer is all about smoothing the social problems in self-assembling communities.
I think the top three would be:
1. Keep it technical. Once things become personal, it’s much harder to have rational conversations.
2. Give everyone the impression you’ve listened to them (however silly the opinion might have been). This is really an extension of keep
it technical – don’t just say no, say no with
a reason.
3. Organise face-to-face meetings at least once a year (possibly in a conference people are already going to). Some of the heat in email discussion seems to happen because people don’t connect the email address at the other end with a real person. Just meeting and talking can help this enormously and remove potential flash points before they erupt.
I suppose as a contribution manager there’s also the secret one… Don’t be too eager to put contributions straight in the tree. Time for reflection helps you (or someone else on the list) see potential problems.

Is SCSI dead? Will it ever be?
So this could be answered in a couple of ways depending on what you mean by SCSI. Traditionally most people use SCSI to mean the SCSI parallel interface (the long ribbon-cabled buses). Hopefully SPI will be dead eventually (say in the next few years), since it already looks like SAS and SATA are taking over (there are a variety of reasons why its passing would be unlamented). If by SCSI, you mean the protocol suite, then no, I can’t really see a point where it’s going to be declared dead. There are signs that the ATA protocols, with version 8 and beyond, are moving in the same direction as SCSI (allowing separation of transports and encapsulations). So in that sense I could see a point where SCSI and ATA essentially become the same protocol, at which juncture ATA zealots could declare SCSI dead, but there will be so much of SCSI in the combined protocol that it won’t really ever be a truthful statement.

Is PPC dead? Will it ever be?
As a competitor to Intel, it’s dead already (just hasn’t stopped moving yet). However, it seems to have several niches outside of Intel’s reach: mainframe, embedded and gaming systems. I can see some of those areas being overrun (Intel is already there in gaming and taking aim at embedded), but it would be a brave person to predict the death of the mainframe (since it’s had so many obituaries by now).

If you could work on any aspect of Linux, what would it be?
Well, I can… that’s the beauty and power of open source: I can pick up and play with any area of it I feel like. Periodically I do dabble in other areas of the kernel. I’ve done the SMP harness for a while, memory management also, and briefly the VFS. None of them really engaged my enthusiasm enough to maintain the contact ongoing, though.
I suspect my next foray will be into the embedded subsystems… seeing if we can get things like Android, which fork the kernel wholesale, back into being respectably close
to mainline.

This article originally appeared in issue 84 of Linux User & Developer magazine.

Linux User & Developer, one of the nation’s favourite Linux and Open Source publications, is now part of the award winning Imagine Publishing family. Readers can subscribe and save more than 30% and receive our exclusive money back guarantee – click here to find out more.