Intel To Increase Engagement With FreeBSD, Makes $250k Donation

[Guest]

butthurt, are we? Anyway, google is your friend. Dell for example is using FreeBSD for enterprise storage. In it's marketed products.

If you go looking, you shall find some more companies using it for various enterprise storage functions.

1 in 100 may be accurate for web servers, latter does not mean "all and every server", of which many are not even connected to the outside world.
World is somewhat wider than your linux, desktop and web servers.

The question is if it's on running bare metal in the case he mentioned. What he said suggests it's just virtual guest, so it doesn't matter what CPU it's running on.

[aht0]

The question is if it's on running bare metal in the case he mentioned. What he said suggests it's just virtual guest, so it doesn't matter what CPU it's running on.

No, the real question is your ideological bias against "anything other than Linux". In the end, it all really comes down to this. Pure bias and bigotry.

[Guest]

No, the real question is your ideological bias against "anything other than Linux". In the end, it all really comes down to this. Pure bias and bigotry.

I can say the same about BSD or proprietary fanboys and developers. When comes to me it's not about being biased. I can point you one by one downsides of BSD and its license.

How big is FreeBSD? The project's Subversion repository is currently about 3.1GB in size; a checked-out tree takes about 600MB. It consists of 71,100 files, about 32 million lines of code. It takes 20-30 minutes to build the whole thing, which is a big improvement from the old days, when it could take several hours.

The project is old, having gotten its start in November 1993; it was based on the 1992 386BSD release which, in turn, was based on much older code. So there is a lot of history. FreeBSD developers have only recently discovered automated testing, he said, but, for such testing to work, the software has to be structured properly, and FreeBSD isn't. Testing the kernel in an automated manner is especially hard. They do have some tests, though, many of which were borrowed from NetBSD, but they are focused mainly on user-space code.

Like most early multiprocessor-capable kernels, FreeBSD had a big kernel lock; Dillon wanted to get rid of it for the FreeBSD 5 release. John Baldwin had been already looking at breaking up the big kernel lock; Dillon joined the effort and things went well for a while. Then the two developers ran into a disagreement over how critical sections should be handled. Baldwin argued for simply disabling interrupts, but Dillon said that was too inefficient; he committed a different change involving a flag in the low-level interrupt handler. It was faster on the x86 architecture, but x86 is not the only architecture supported by FreeBSD.

This change led to a number of big arguments, with Dillon eventually backing out the change. A month-long acrimonious debate followed; eventually Dillon's change went in, but there were a lot of other, similar incidents. It was a classic case of a "rock-star ninja 10x developer" not getting along with the rest of the project.

Other problems are tied to the project's age, but that is hard to fix too. FreeBSD will not be getting any younger. It can be hard to change baked-in attitudes. For example, there is resistance to moving to GitHub because people in the core team and beyond are attached to running their own infrastructure. Such things can be worked around, but an attitude shift will be needed.

With regard to leadership, that could be fixable but it, too, would require an attitude shift in the core team. The team does not need to be an architectural leader, it needs to be a "scaffolding leader". It just needs to ensure that "the things the project needs to do a good job" are available. The core team should focus on how FreeBSD is made, rather than what it should be. Its job should be reducing friction for developers.

But that points to the real trouble with FreeBSD: it is made and led by volunteers. There is nobody whose job is purely to make FreeBSD more awesome. There are plenty of people tasked with making it more awesome for their employers, but that work doesn't always help in the long term. Those employers get a faster build system, but automated testing languishes. Somehow, the project needs more resources to be able to focus on making FreeBSD "a fun and awesome place" because that is how they can ensure that it will be around for a long time to come.

https://lwn.net/Articles/712308/

[aht0]

Regardless how you go about excusing your attitudes or whatever you may link for whatever reason - you have been displaying the attitude of "I hate, attack and insult everything that is not Linux, Linux-related or somehow competing with a Linux.".

You have been doing it consistently and without slightest variation - attacking BSD's, Solaris, Windows and whatever else happens to cross your path. Usually leaving clear impression of plain hatred and showing very little reason. Once some new BSD/Solaris thread shows up in the forum, especially comparison testing with some Linux, you can guarantee mr.Pawlerson is there spreading sap and insults on all directions.

You, are, in an essence, an hypocrite idiot. You are at once imagining yourself standing for software freedom (Linux) but at the same time attacking diversity of free software. It's what defines the "bigot".