Slashdot Mirror
Tridge Speaks Out
Robert McMillan from Linux Magazine posted an interview with Tridge, of Samba and Tivo fame. He's one of the most important folks in all of Linux, and this interview is worth a read. He covers a lot of good material like crap code, bonobo, and what stuff in the kernel is innovative. He also talks a bit about what he might do after Microsoft drops SMB from future versions of windows.
from the article:
" . . . (who recently jumped ship from Linuxcare to work at VA Linux Systems) . . ."
this guy has made some stunningly bad career choices. I hope, for his sakes, Australia has a decent welfare system. Sheesh!
Slashdot, come for the goatse, stay for the trolls.
You end up exchanging dozens of e-mails, where you say, "That's bad because of this and this," and they say, "Oh no, this programming style is great." Then you have got to teach them a couple of years of computer science so they can understand why it's crap.
As far as I know VA I.O.U. eliminated all their programmers, including the Australian group. Conspicuously absent from the interview is #1 whether he's employed and whether open source projects contributed anything positive to his employment prospects and #2 how he's handled liability for the use of Samba.
Why didn't they ask him about tivo? They mention his Tivo hacking in the into but then, leave the readers hanging.
Is here.
Hmm, you'd think a "printer friendly" version would remove JPGs of Unix-nerds. No one wants to print those out. Actually, I'd be happy if that were the only thing it did.
Athletic Scholarships to universities make as much sense as academic scholarships to sports teams.
All the interesting positions in Linux are taken?
In the kernel, perhaps, and low-level network services. The barrier to entry is high there, and for the best possible reasons. The OS has become a phenomenon - and very very bright people continue to contribute. Lesser lights need not apply until they're more seasoned.
But that doesn't mean that there are applications out there. Many beautiful tools exist - but truly widespread adoption of Linux will require not just killer applications, but also many, many mundane little applications to fill all the niches where little Windows apps fit.
We all know about StarOffice/OpenOffice. But. . .
A really nice accounting package? A little one, like QuickBooks? Contact managers like GoldMine? Bigger CRM applications with all the bells and whistles?
These are projects that are begging to worked on, and completed. (And yes, I know that good work is being done on all these fronts already).
And these are just the sorts of projects that would convince people (at my company at least) to ditch Windows on the desktop once and for all. Even management is becoming painfully aware of Microsoft's hardball tactics, with this latest round of forced upgrades.
There are plenty of crucial positions left in Linux development. Only the shift is now moving away from the kernel and services, and toward real business applications.
Truly an exciting time. And we're only in the 2nd inning of the revolution.
If you're a company that is trying to compete directly against a free product, you're in trouble.
Now I understand why all those Linux companies went bankrupt. They were competing against each other!
Actually, I would like to have asked him if he still receives occasional "pizza payments" for Samba.
Don't know if it still says it, but in the old docs for Samba, they used to have an address where you could FedEx pizza donations to the Samba team.
Now *that's* my idea of compensation. Free as in beer? Well, what goes better with free beer than free pizza!
If it ain't broke, it doesn't have enough features yet.
Forget all the other replies to this post. The entire idea of Linux and it's You-Want-It-You-Build-It approach is very silly. If I want a tool that I can't build, I can certainly encourage others to make it through other means of assistance, IE - software/hardware donations, monetary donations, other project investments and involvement...
Linux is being held back not only by its lack of quality offerings in a few software genres (Such as the aforementioned QuickBooks), but also by the utter *glut* of offerings in some of the other areas. It's really daunting for some people that most Linux distributions come on five or six CDs and Windows only comes on one. I thought distributions were supposed to be like customized toolsets for different people and different situations. If this is the case, then why do all the distributions have almost all the same software?
Hear hear, man. Mod parent UP.
I must say that Andrew seems to be very aware of Samba's direction and future. He also seems to be very honest about the delevopment-to-date. I really liked how he points out the learning phases, from concept to write to rewrite that Samba has taken over the years. My only concern would be that as awesome as Samba is, that Microsoft would dump the SMB protocol completely, rendering the current implementations useless with newer Microsoft OS'. Right now, my employer (like most other employers) use Windows on its servers and desktops, the only exceptions are the CAD designer's that I and my buddy take care of, but we have Samba running on our Sun E450, authenticating to our Windows PDC and it works great. It be really suck if they upgrade our Windows servers to a newer version that doesn't support SMB, but that would just be the next challenge to software developers to once again try and accomadate Microsoft, and once again prove that no matter what Microsoft does, hackers will always find a way. Hooray for Andrew and all hackers like him.
One word: AutoCAD.
(Someone showed me a non-Free, relatively cheap CAD package for Linux, but I'm suggesting a GPL equivalent).
Also, you'll get some pretty heavy heat on the style issue. Casting malloc(), using redundant parantheses on sizeof (i.e. treating it as a function or macro), performing micro-operations (which the compiler/interpreter does a much better job of) are all grounds for assault.
comp.lang.c will set you straight within a couple of weeks (given you're willingness to be humiliated on a regular basis), and you'll be still be learning things after years of regular reading. Again: comp.lang.c is very strict: if comp.lang.c++ is Woodstock, comp.lang.c is a Nazi concentration camp. Okay maybe I'm overselling it a bit
As for automatic code checkers, use lclint liberally. There's also the mythical Stanford patch for gcc (at least I think it's from Stanford; one of those pretentious schools anyway) that was used to find many Linux bugs.
Seriously, just start posting random tiny (no longer than 3 or 4 small functions, or else people will get bored of reading) snippets of code to comp.lang.c. 95% of posters can't even get a satisfactory Hello World! on their first try.
Those are not fun. Amateurs/hackers work on fun things. If you want non-fun things, you have to pay someone to do it.
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
the interview was in the july issue (as the bottom of each page in the linked article indicates). the interview itself could have taken place a month or two earlier than that. So likely he was still employed then.
News for Geeks in Austin, TX
Depends on your definition of fun. Personally I get a lot of satisfaction/enjoyment/warm-fuzzies from building stuff that people use.
As i'm sure others do too.
If enough people make enough noise about wating package X, someone will go write it. I promise.
My only concern would be that as awesome as Samba is, that Microsoft would dump the SMB protocol completely, rendering the current implementations useless with newer Microsoft OS
I don't think they CAN make it completely unusable with previous versions of SMB or else no previous Windows products could work with it either, unless you're forced to download a new networking patch for EVERY Windows PC on your network. Course, if it was installed via the next IE update, then it might work...
If previous Windows clients can't talk to the new stuff, people will think VERY long about upgrading and having to throw away all the current investment.
creation science book
I think the problem is a little different than the project being "fun" or not.
Me? My team and I built and maintain a homegrown oracle-based solution that stitches together contact management, accounting, and production management for a little manufacturing firm here in Chicago.
I find these things... fun.
The problem is - it takes a lot of expertise and insight (garnered only through painful experience) to build applications heavy in accounting and business-logic functions . . . expertise that goes beyond what is normally taught in a CS program in school.
And recruiting people with expertise outside the computer science realm into open source development (without paying them - your point is well taken here) can be really difficult.
There just aren't that many people with cross-disciplinary expertise who aren't buried in paying-job-type work and who are willing to donate to the cause.
What will no doubt continue to happen is new versions of SMB that are negotiated between machines if both support them. That DOS 3.3 client winds up talking a much simpler version of SMB than a Windows 2000 client would negotiate with a Windows 2000 server. So eventually you discover that your old clients don't support some fancy new feature or performance enhancement. But they should continue to work as they have.
When I worked at Microsoft there were stories that SMB would be dropped in favor of something else. One of these was the famous "HTTP redirector", which would be a client that connected to a web server to replace SMB. One big advantage of HTTP is that it allows a file to be read with one single network roundtrip. SMB separates out the operations of open, read, and close a file (those take up 3 separate SMBs). Now the protocol also allows SMBs to be combined in a single request, but various cruftiness that have built up in the protocol prevents you from actually packaging an open, read, and close in one packet and getting a response back with all the data in one roundtrip. This "small file read" test is a favorite of labs doing performance tests, so although HTTP is slightly larger and slower to parse, that all winds up being noise compared to doing the whole small file read in one roundtrip.
Unfortunately HTTP has some *disadvantages*, in particular a lot of the other semantics supported in SMB, like printing and locks and various security protocols, are not supported, so those would have to be invented. The push for an HTTP redirector seemed to be coming from the Internet Explorer and IIS teams (I worked on NT). In fact one IIS person confidently told me about 4 years ago that SMB was doomed. But it continues to live on and the various HTTP redirector projects seem to have stalled. But for all I know it is still bubbling around somewhere (or several places) at Microsoft.
- adam
From the interview:
."
AT: It depends how related it is. If Microsoft does a completely new protocol, it would make far more sense to have a second package. Whether I do that package or not depends on how stupid I feel at the time.
LM: How stupid do you feel right now?
BWAHAHAHA!!
" . . . (who recently jumped ship from Linuxcare to work at VA Linux Systems) . .
I wanted his answer to be, " really stupid considering the companies I have worked for lately."
Nah, can't get that lucky.
ACK
If any project deserves developers, it's this one.
Life's a bitch but somebody's gotta do it.
He marked some of mine when I took his Operating Systems Unit at Uni.
How I supposed to know you shouldn't use 'float' in the kernel. The code worked, that was good enough for me. Still not 100% sure I know why this is the case.
Everything I am today I owe to people, whom it is now to late to punish."
ms introduced port 445, 'microsoft-ds', with win2k. one can completely disable netbios on a system and still transfer files, but network neighborhood won't work anymore - it's going to be a short amount of time before there's something to replace that.
Not quite. The 445 stuff is actually SMB. It's just not over NetBIOS. That little 4 byte header is left blank (or maybe it encodes a length, I don't remember exactly). And Network Neighborhood (aka Browsing) is replaced with similar functionality using Active Directory I beleive. Later versions of Samba support the CIFS on port 445 I beleive.
Okay, this is so far the second place I've seen this mentioned. Does anyone have any more info on Microsoft dropping SMB? Info on what they'll allegedly replace it with? I haven't been able to dig up any real dirt on this, and I haven't gotten more than hearsay, this snippet here and a mention in passing by an Apple sales guy. Any info would be helpful.
.. etc. It would be quite a headache to decouple all of that stuff. I have not heard *anything* about ditching it entirely. It would immediatedly render the default install of any of their OSs useless with the new product. Besides, there is a fairly substantial CIFS community at the moment. You have EMC, Netapp, Unisys, and others. Not gonna happen soon.
I don't think the're going to get that monkey off their back for a while. There's a lot of stuff funneled through CIFS (aka SMB). You've got NamedPipes, RAP calls, transactions, DCE/RPC,