Bill Manning has died | Hacker News Reader

Bill Manning has died

rdvlivefromtokyo.blogspot.com

119 pointsdredmorbius67 comments

For anyone wondering who he was or his contributions:

> Manning has been working in the Internet industry since 1979 when he started working at Texas Instruments and helped in building its IP network. After which he joined Rice University and made SESQUINET. He played a significant role in the migration of MIDNET and SESQUINET from NSFnet regional networks to commercial networks.

> He worked on the COREN and CALREN-2 technical committees. At ISI he worked in the Routing Arbiter Project.

> Bill has been working with the the IETF and IEPG as an individual participant, working group chair, and code developer. He specfied the method to add NSAP support to the DNS.

https://icannwiki.org/Bill_Manning

Thanks for that.

I always feel bad seeing that someone has died and how important they were in computer science but it's the first time I've heard of them.

It's sad that when you google him and you only get results for a sports executive.

He was also a DNS root server operator for many years (he ran b.root-servers.net)

When I see eulogies like this for people I've never heard of, I'm reminded sometimes that everything good in my life that I take for granted exists because of someone somewhere just doing the best they can. It helps ease the angst.

I knew Bill professionally, he was a great guy and we'll all be at a loss without him around.

Back in 2003 when I started SFIX he was instrumental in getting it off the ground.

I can't believe he's gone.

I'm sorry for your loss.

Some stand-up comic in the 90's had a joke to the effect of:

And when you hit a certain age, your friends start playing a game called, "Guess Who Died?"

This seems like the year when all the DotCom era kids start playing the game.

Flipside is that many of the founding fathers of computers, software, networking, and security are still alive, and can still tell their stories.

That's an opportunity for any historians out there.

One of the best experiences of my college life was escorting Tony Hoare around the campus during his several talks that day (as well as attending those talks of course). I distinctly remember thinking how lucky I was to be around during the time of the giants of my field. Like a physics major getting to sit and talk with Newton or similar.

> On other occasions, when the waiter asked for his order, Bill would point to another person at the table, and say, "I'll have what she's having." "Well, what is she having?" "I don't know, I haven't heard her say." Once in a while, he would point to someone else in the restaurant and say, "I'll have what they are having." It was funny and sometimes disconcerting, which was very Bill, and it was also his way of making sure he himself was eating (and thinking and doing) as broadly as possible, without getting stale.

What an endearing fellow. I wish I was more like him.

I'm always impressed (and admittedly envious) when I see a person who manages to get a PhD while skipping the bachelors degree; being a dropout I wish I could do that, but of course I didn't build TI's IP network or SESQUINET :)

RIP Bill; HN should consider putting a black bar.

Does anybody know more about how this happens? As a fellow dropout, I've sometimes wanted to go back for a master's, but there's no way I'm spending n years getting a bachelor's first.

Could be a ton of fun tho.

From the research I've done on this, it looks like you typically have to have made huge contributions to a field already, and have a professor or ten advocating for you at a university to accept you into the PhD program.

I've emailed about twenty of math professors in the NYC area to let me into a masters or PhD program, since I thought my work history for the big "brand name" compsci research companies and my work as an ostensible research scientist a a big nice university would suffice, but sadly not a single one was even willing to meet with me.

Knowledge > credentials. Hope it doesn’t get you too down, sometimes the only winning move is to not play.

You're not wrong, though it does sadly preclude me from the more interesting compsci/math research positions (stuff involving type theory and compilers and whatnot); most universities and research teams (understandably) have a firm minimum requirement of a masters degrees, preferring a PhD.

I totally understand why they do this...if you're dedicated enough to get a PhD in a field then you're almost-inherently going to be a decent researcher, so I can't blame whomever is in charge of hiring at places like MS Research for having these criteria. It's just a bit sad that this means, at minimum, even if I were an otherwise perfect candidate, I probably couldn't get a job doing real research for at least another seven years (3 to finish the bachelors, and 4 to get the PhD if I'm being very generous).

EDIT: To those wondering if this is contradictory to my previous comment, I will clarify; I do work for one of the big brand name compsci research companies, but (sadly) not doing research. When I worked for a big university, my title was technically "research scientist", but my actual job could have been more-accurately described as "code-monkey", and that's putting it generously.

Never give up on applying for roles you believe you’re qualified for even though you’re missing some letters, every once in a while someone breaks the mold and hires based on the right signals (versus a checkbox) in this scenario. Applying is free, you miss 100 percent of the shots you don’t take.

Oh, I do keep applying, I just try to not get me hopes up.

I think Gregor Kiczales mentioned having no undergrad degree -- this was from a conversation in the 90s, so I have low confidence in the memory. IIRC he worked on CLOS and its metaobject protocol at PARC before going into academia. (I was looking to join his group as another guy without a degree.)

It might be harder now.

Maybe there is a little bit of solace in knowing that in some sense he lived more in his too short life then most of us will in our full-length ones.

RIP

Re: ...we called Bill "the bad idea fairy". He always brought a slightly-off-kilter view of technical problems, which triggered endless discussions of fascinating, if usually implausible, alternatives.

I've been "accused" of similar. Although, I can't claim I'm as successful as Bill. I tend to ruffle sacred cows. It's not like I'm trying to agitate people, I just view the world a bit different from established opinions.

For example, we have dynamic programming languages, but not dynamic RDBMS (that use SQL or a close variation). Why not?

And we have XML as a fairly flexible meta-standard for data structures, but why not something comparable for C-like-syntax programming languages? You could roll your own programming language without creating a parser from scratch (and hopefully mix and match behavior based on "part kits").

Both of these seem like logical extensions of existing tools to me, but I get a lot of vague flack. It's like the universe is calling for them, not just me, yet nobody cares. It's not outright inventing new concepts, it's just taking concepts that worked well in one computing domain and applying them to another. They might fail, but so might every other IT experiment going on currently.

> For example, we have dynamic programming languages, but not dynamic RDBMS (that use SQL or a close variation). Why not?

Depending on what you mean, SQLite is exactly this.

It is relational, so one does have to structure tables, but I consider this equivalent to naming variables.

Unlike most (all?) other SQL databases, however, the 'types' of the data are just suggestions, and have some influence on how SQLite stores and retrieves them internally.

But you can put a string in an INTEGER field, SQLite won't stop you.

But you have to pre-specify the columns. In a truly dynamic DB, you could enter "INSERT INTO foo (myNewColumn) VALUES (123)" and both the table "foo" and "myNewColumn" are created on the fly. SqlLite doesn't do this. Further, one could add restrictions as needed to "lock down" both new column creation, and add type-like validation ("parse" check). Thus, as requirements solidify, it can act more like a traditional RDBMS.

> For example, we have dynamic programming languages, but not dynamic RDBMS (that use SQL or a close variation). Why not?

JSON datatypes essentially serve that need nowadays. Postgres even lets you write queries against them.

Generally, a lot of the value from RDBMS in fact comes from strong guarantees that you can make about the data you're storing (datatypes, constraints, unique indexes, etc). That's what makes them inherently better than MongoDB or whatever. But if you really just want a "misc" column (or to use PG as a key-value-store) then you can do that with a JSON column.

What use-case are you looking at that wouldn't be solved by a JSON column?

> And we have XML as a fairly flexible meta-standard for data structures, but why not something comparable for C-like-syntax programming languages? You could roll your own programming language without creating a parser from scratch (and hopefully mix and match behavior based on "part kits").

Well, isn't that lex and yacc?

Re: JSON datatypes essentially serve that need nowadays. Postgres even lets you write queries against them.

Yes, but JSON columns are treated mostly different than regular columns. They are like second-class citizens that require more verbosity to work with. Dynamic Relational won't have this dichotomy. Columns are columns.

Re: Well, isn't that lex and yacc?

No. You could make a parser for C, Lisp, Fortran, Datalog, etc. with them. They don't offer a standard. I'm proposing a standard syntax or meta syntax. Then the common existing languages could offer a parser for this meta language similar to how they offer an XML parser. You don't have to build the XML parser yourself nor give it the parsing rules for XML, like you have to with lex/yacc. You couldn't do that with lex and yacc in a practical way, they are too big and open-ended.

Re: a lot of the value from RDBMS in fact comes from strong guarantees that you can make about the data you're storing

There is a time and place for static and a time and place for dynamic. Further, Dynamic Relational can incrementally apply constraints to "lock down" various aspects, as I mentioned nearby. For example, you could add a "parse check" to a given column to ensure its parse-able as a integer upon insert and/or updates. Thus, out of the box it's loosey goosey, but doesn't have to stay that way.

I'm working on a write-up to explain this all with SQL examples.

Re: mixing and matching parts, you are describing Lisp. And probably Racket.

They are not "C-like" in their syntax, at least in an enforced kind of way. (What I proposed had to deviate from C-like a bit to ensure NON-reliance on key-words to know where block-groups start and end. I had to find a middle ground between being "meta" and being C-like. I nicknamed it "moth" based on the shape of the primary "atomic" structure.)

> And we have XML as a fairly flexible meta-standard for data structures

Are you a time traveler from 15 years ago? :-)

More often than not those 'bad ideas' lead with a bit of twisting to good ideas. I've seen quite successful companies started like this multiple times, and it is an essential component in a good pivot.

Really tragic. RIP.

Reply on news.ycombinator.com