I have long advocated that somehow the companies that have built their businesses on the backs of open source developers find some way to reward them on the backs of the IPO, like redhat did, in 1999. This is not that. For a one time investment in stock, you too, can support development of our new AI overlords.

This is why I trust my (cerowrt, now openwrt) wndr3800s far more than any modern PC or embedded hardware. No blobs. I sleep better. There were only two major CVEs I was ever worried about - one was kind of big, and in the FPU emulator...

I keep hoping someone will give me a simple packet capture of a single 1 minute tcp download via fq_codel shaping inbound on pfsense. I am pretty sure at this point there is something wrong here, but lack the skills and time to sort out the complaints from the reality. https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.reddit.com%2Fr%2Fopnse...

I still use venerable WNDR3800s (15 years old now) as APs. (I had about 30 left over after the make-wifi-fast project ceased) They do 300Mbit, no binary blobs, have good range, stay up forever (I know of people with 3+years uptime), and are the best known fq_codel implementation across the board.

Elsewhere I kind of gave up on an all-in-one unit for gbit+ networking and went with the evenroute pro (sadly deceased, but the company was VERY good about upgrading their userbase to mainline OpenWrt) - but any x8664 mini-pc with few ethernet ports suffices nowadays.

The mt76 and mt79 wifi chips stablized a lot in the past year (and have way less "blob" to them than the qcomm gear - so they are looking like the successor to the ath9k for me.

fq_codel native on the mt76 and mt79 chips is the bomb. https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fblog.cerowrt.org%2Fpost%2F... OpenWrt has CAKE also. I am seeing a lot of *sense fanbois complaining that fq_codel shaping inbound on BSD is seemingly buggy, and those that went from opnsense to OpenWrt, much happier with CAKE on the QoE front. I have been trying to find someone with BSD experience for ages to help figure out what is going wrong on that OS in this department.

I wish t-mobile, etc, would realize that if they fixed their bufferbloat, the complaints about speed would vanish, and they could still throttle big users to their hearts content. fq_codeled 5G in my lifetime would be nice to see.

I remember that now, thx!

Borland I was working on the interbase database. Zero comms with the compiler team. The gcc patches I attempted to get out were to get our (Unix) code built on some obscure platform (this was 1992). Ixnayed. Any sufficiently big software company will have some potential conflict with open sourcing anything. Most recently, the authors of RFC9406 - an attempt at an open document, prohibited from working directly on a Linux implementation.

SCO did not want anyone to work on Linux. In fact, got a restraining order against me or 3 years against working on multiprocessor systems dSybase did not want anyone working on postgres Mediaplex did not want anyone contributing to apache. That was my life 1992-2000. I am glad your life was better. Android consciously ripped out all gplv2 code from userspace Google attempted to get dnsmasq relicensed. Google was apache only by default. 2018 apple would not permit gplv3 code Today gplv3 is essentially banned everywhere I have worked. OpenWrt derivatives - Comcast rdk-B, and a zillion corps that use it, rip out all the gpl components and essentially prohibit their devs from contributing back to it.

Does anyone remember Apple's attempt to rewrite their OS from scratch using modern C++ techniques? They burned a decade on it. https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fen.wikipedia.org%2Fwiki%2F...

Linux ate the Unix server market, as well as a goodly portion of the windows server market.

If I could mod that comment up 100, I would.

Some discussion of the Kyle question on esr's site here: https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fesr.ibiblio.org%2F%3Fp%3D875... - the cert is invalid now, but you can click through to 888+ comments there.

I remember picketing frys and handing out CDs of Linux, and protesting to stop paying the Microsoft Tax!
I remember being afraid for my job if I contributed to FOSS under my own name.
I remember the real-world evidence that had accumulated that Linux + Samba was massively better than NT + SMB - I had had a site that crashed 3 times a week with NT, and replaced it with Linux.
I remember the ease of remote support for that, leveraging ssh rather than a gui.
I remember how we as a community pulled together to tackle many of the real problems we had had then (like multi-processor support) support that Microsoft had identified for us.
In looking over that old Microsoft strategy today - I do see many bothersome things - like "decommoditizing protocols" - that still infect the industry. It would be good for more here to re-read the first and second Halloween documents and reflect on the good and the bad!
One thing entirely missed by all in 1998 is the rise of the web (and phone!) replacing applications that ran locally, with things like drag and drop. Sometimes I point to the GPLv2 as being a proximate cause of the rise of the cloud as that kept the custom code out of the customer hands.

Thx for reposting this.Very few modern Linux folk seem to remember the context of the original war. What will the next 25 years look like?