I've been too busy with interviews (and my 24 Hours of LeMons race
over the weekend) to prepare for the class, so I'm going to have to
push off the NIO fun stuff for yet another week. Sorry folks... I'll
get there. :)
marsh...@software.mpierce.org> wrote: > I've been too busy with interviews (and my 24 Hours of LeMons race > over the weekend) to prepare for the class, so I'm going to have to > push off the NIO fun stuff for yet another week. Sorry folks... I'll > get there. :)
The bad news: still no class on Tuesday -- I don't have enough prepared on the NIO project to take up a full class. Also, the following week I'll be out of town and I won't be back until the 29th, so that means the next time I'll be in town will be January 4th. I hate to not have class for that long, so if anyone has ideas on something they'd like to present in my absence, please propose them!
The good news: I'm finally done with the interviewing/job hunt process as of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm stunned at how much time it took... I guess because it wasn't associated with being in an office, it never was really done for the day.
Speaking of the NIO project, I've got a cool (and useful) project for us to work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with IPV6 for various reasons. I think it would be interesting to make a DNS server implementation that (1) works with IPV6 and (2) is as scalable as we can make it. Such a server would need to handle a crazy amount of requests/sec, but each request is just a UDP packet (no TCP connections!). Conveniently, this means we get to focus on scalability without worrying too much about the annoying parts of keeping track of connection state. Naturally, we would also need to figure out a way to store the information that the server would serve to make it useful, but to start with it's just a neat real-world application of NIO.
marsh...@software.mpierce.org> wrote: > The bad news: still no class on Tuesday -- I don't have enough prepared on > the NIO project to take up a full class. Also, the following week I'll be > out of town and I won't be back until the 29th, so that means the next time > I'll be in town will be January 4th. I hate to not have class for that long, > so if anyone has ideas on something they'd like to present in my absence, > please propose them!
> The good news: I'm finally done with the interviewing/job hunt process as > of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm > stunned at how much time it took... I guess because it wasn't associated > with being in an office, it never was really done for the day.
> Speaking of the NIO project, I've got a cool (and useful) project for us to > work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with > IPV6 for various reasons. I think it would be interesting to make a DNS > server implementation that (1) works with IPV6 and (2) is as scalable as we > can make it. Such a server would need to handle a crazy amount of > requests/sec, but each request is just a UDP packet (no TCP connections!). > Conveniently, this means we get to focus on scalability without worrying too > much about the annoying parts of keeping track of connection state. > Naturally, we would also need to figure out a way to store the information > that the server would serve to make it useful, but to start with it's just a > neat real-world application of NIO.
> Thank You Marshall: the delay is perfect. Feel free to delay to the second > week in Jan. Still haven't finished the fork-join homework.
> Great idea on the NIO DNS server!!! I think this is my favorite class of > all time! > dc
> On Mon, Dec 13, 2010 at 11:27 PM, Marshall Pierce < > marsh...@software.mpierce.org> wrote:
>> The bad news: still no class on Tuesday -- I don't have enough prepared on >> the NIO project to take up a full class. Also, the following week I'll be >> out of town and I won't be back until the 29th, so that means the next time >> I'll be in town will be January 4th. I hate to not have class for that long, >> so if anyone has ideas on something they'd like to present in my absence, >> please propose them!
>> The good news: I'm finally done with the interviewing/job hunt process as >> of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm >> stunned at how much time it took... I guess because it wasn't associated >> with being in an office, it never was really done for the day.
>> Speaking of the NIO project, I've got a cool (and useful) project for us >> to work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work >> with IPV6 for various reasons. I think it would be interesting to make a DNS >> server implementation that (1) works with IPV6 and (2) is as scalable as we >> can make it. Such a server would need to handle a crazy amount of >> requests/sec, but each request is just a UDP packet (no TCP connections!). >> Conveniently, this means we get to focus on scalability without worrying too >> much about the annoying parts of keeping track of connection state. >> Naturally, we would also need to figure out a way to store the information >> that the server would serve to make it useful, but to start with it's just a >> neat real-world application of NIO.
I'm finally back in California -- had a lot of fun with airlines over
the holidays. Nothing like being on vacation without your luggage.
I'm trying to get everything together for a class on Jan 11th. Lot of
work still to do on the NIO project -- see the aforementioned note
about lack of luggage. :) I'll also have some coverage of Scala
actors. I looked into Groovy as well during the break, but the
language doesn't impress me. If there's interest, though, I could do a
little bit on gpars.
On Dec 14 2010, 2:17 am, Doug Chang <doug.ch...@hackerdojo.com> wrote:
> Thank You Marshall: the delay is perfect. Feel free to delay to the second
> week in Jan. Still haven't finished the fork-join homework.
> Great idea on the NIO DNS server!!! I think this is my favorite class of all
> time!
> dc
> On Mon, Dec 13, 2010 at 11:27 PM, Marshall Pierce <
> marsh...@software.mpierce.org> wrote:
> > The bad news: still no class on Tuesday -- I don't have enough prepared on
> > the NIO project to take up a full class. Also, the following week I'll be
> > out of town and I won't be back until the 29th, so that means the next time
> > I'll be in town will be January 4th. I hate to not have class for that long,
> > so if anyone has ideas on something they'd like to present in my absence,
> > please propose them!
> > The good news: I'm finally done with the interviewing/job hunt process as
> > of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm
> > stunned at how much time it took... I guess because it wasn't associated
> > with being in an office, it never was really done for the day.
> > Speaking of the NIO project, I've got a cool (and useful) project for us to
> > work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with
> > IPV6 for various reasons. I think it would be interesting to make a DNS
> > server implementation that (1) works with IPV6 and (2) is as scalable as we
> > can make it. Such a server would need to handle a crazy amount of
> > requests/sec, but each request is just a UDP packet (no TCP connections!).
> > Conveniently, this means we get to focus on scalability without worrying too
> > much about the annoying parts of keeping track of connection state.
> > Naturally, we would also need to figure out a way to store the information
> > that the server would serve to make it useful, but to start with it's just a
> > neat real-world application of NIO.
Welcome back!
Didn't realize that will see you guys a year later since the last
class :-)
Very looking forward the NIO class, whose idea? I owe him a thank
you :-).
Regarding the new language, James Gosling said he will choose Scala if
Java dies(at his Java talk @Google in Nov).
library is easy, the runtime is tough. In near future, he said he
wasn't worry anything can replace Java.
Well, if Scala uses Java runtime, then it's possible will be wildly
adopted, just my 2 cents.
~Jing
On Jan 3, 1:41 pm, Marshall Pierce <marsh...@software.mpierce.org>
wrote:
> I'm finally back in California -- had a lot of fun with airlines over
> the holidays. Nothing like being on vacation without your luggage.
> I'm trying to get everything together for a class on Jan 11th. Lot of
> work still to do on the NIO project -- see the aforementioned note
> about lack of luggage. :) I'll also have some coverage of Scala
> actors. I looked into Groovy as well during the break, but the
> language doesn't impress me. If there's interest, though, I could do a
> little bit on gpars.
> On Dec 14 2010, 2:17 am, Doug Chang <doug.ch...@hackerdojo.com> wrote:
> > Thank You Marshall: the delay is perfect. Feel free to delay to the second
> > week in Jan. Still haven't finished the fork-join homework.
> > Great idea on the NIO DNS server!!! I think this is my favorite class of all
> > time!
> > dc
> > On Mon, Dec 13, 2010 at 11:27 PM, Marshall Pierce <
> > marsh...@software.mpierce.org> wrote:
> > > The bad news: still no class on Tuesday -- I don't have enough prepared on
> > > the NIO project to take up a full class. Also, the following week I'll be
> > > out of town and I won't be back until the 29th, so that means the next time
> > > I'll be in town will be January 4th. I hate to not have class for that long,
> > > so if anyone has ideas on something they'd like to present in my absence,
> > > please propose them!
> > > The good news: I'm finally done with the interviewing/job hunt process as
> > > of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm
> > > stunned at how much time it took... I guess because it wasn't associated
> > > with being in an office, it never was really done for the day.
> > > Speaking of the NIO project, I've got a cool (and useful) project for us to
> > > work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with
> > > IPV6 for various reasons. I think it would be interesting to make a DNS
> > > server implementation that (1) works with IPV6 and (2) is as scalable as we
> > > can make it. Such a server would need to handle a crazy amount of
> > > requests/sec, but each request is just a UDP packet (no TCP connections!).
> > > Conveniently, this means we get to focus on scalability without worrying too
> > > much about the annoying parts of keeping track of connection state.
> > > Naturally, we would also need to figure out a way to store the information
> > > that the server would serve to make it useful, but to start with it's just a
> > > neat real-world application of NIO.
> I'm finally back in California -- had a lot of fun with airlines over
> the holidays. Nothing like being on vacation without your luggage.
> I'm trying to get everything together for a class on Jan 11th. Lot of
> work still to do on the NIO project -- see the aforementioned note
> about lack of luggage. :) I'll also have some coverage of Scala
> actors. I looked into Groovy as well during the break, but the
> language doesn't impress me. If there's interest, though, I could do a
> little bit on gpars.
> On Dec 14 2010, 2:17 am, Doug Chang <doug.ch...@hackerdojo.com> wrote:
> > Thank You Marshall: the delay is perfect. Feel free to delay to the second
> > week in Jan. Still haven't finished the fork-join homework.
> > Great idea on the NIO DNS server!!! I think this is my favorite class of all
> > time!
> > dc
> > On Mon, Dec 13, 2010 at 11:27 PM, Marshall Pierce <
> > marsh...@software.mpierce.org> wrote:
> > > The bad news: still no class on Tuesday -- I don't have enough prepared on
> > > the NIO project to take up a full class. Also, the following week I'll be
> > > out of town and I won't be back until the 29th, so that means the next time
> > > I'll be in town will be January 4th. I hate to not have class for that long,
> > > so if anyone has ideas on something they'd like to present in my absence,
> > > please propose them!
> > > The good news: I'm finally done with the interviewing/job hunt process as
> > > of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm
> > > stunned at how much time it took... I guess because it wasn't associated
> > > with being in an office, it never was really done for the day.
> > > Speaking of the NIO project, I've got a cool (and useful) project for us to
> > > work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with
> > > IPV6 for various reasons. I think it would be interesting to make a DNS
> > > server implementation that (1) works with IPV6 and (2) is as scalable as we
> > > can make it. Such a server would need to handle a crazy amount of
> > > requests/sec, but each request is just a UDP packet (no TCP connections!).
> > > Conveniently, this means we get to focus on scalability without worrying too
> > > much about the annoying parts of keeping track of connection state.
> > > Naturally, we would also need to figure out a way to store the information
> > > that the server would serve to make it useful, but to start with it's just a
> > > neat real-world application of NIO.
Yeah, I think TCP is the way to go. There are some interesting things
about reading from local storage but I get the feeling more people are
interested in networking.
> Marshall,
> I assume you will teach/or touch a lot TCP/IP socket/streaming while
> you teach NIO?
> ~Jing
> On Jan 3, 1:41 pm, Marshall Pierce <marsh...@software.mpierce.org>
> wrote:
> > I'm finally back in California -- had a lot of fun with airlines over
> > the holidays. Nothing like being on vacation without your luggage.
> > I'm trying to get everything together for a class on Jan 11th. Lot of
> > work still to do on the NIO project -- see the aforementioned note
> > about lack of luggage. :) I'll also have some coverage of Scala
> > actors. I looked into Groovy as well during the break, but the
> > language doesn't impress me. If there's interest, though, I could do a
> > little bit on gpars.
> > On Dec 14 2010, 2:17 am, Doug Chang <doug.ch...@hackerdojo.com> wrote:
> > > Thank You Marshall: the delay is perfect. Feel free to delay to the second
> > > week in Jan. Still haven't finished the fork-join homework.
> > > Great idea on the NIO DNS server!!! I think this is my favorite class of all
> > > time!
> > > dc
> > > On Mon, Dec 13, 2010 at 11:27 PM, Marshall Pierce <
> > > marsh...@software.mpierce.org> wrote:
> > > > The bad news: still no class on Tuesday -- I don't have enough prepared on
> > > > the NIO project to take up a full class. Also, the following week I'll be
> > > > out of town and I won't be back until the 29th, so that means the next time
> > > > I'll be in town will be January 4th. I hate to not have class for that long,
> > > > so if anyone has ideas on something they'd like to present in my absence,
> > > > please propose them!
> > > > The good news: I'm finally done with the interviewing/job hunt process as
> > > > of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm
> > > > stunned at how much time it took... I guess because it wasn't associated
> > > > with being in an office, it never was really done for the day.
> > > > Speaking of the NIO project, I've got a cool (and useful) project for us to
> > > > work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with
> > > > IPV6 for various reasons. I think it would be interesting to make a DNS
> > > > server implementation that (1) works with IPV6 and (2) is as scalable as we
> > > > can make it. Such a server would need to handle a crazy amount of
> > > > requests/sec, but each request is just a UDP packet (no TCP connections!).
> > > > Conveniently, this means we get to focus on scalability without worrying too
> > > > much about the annoying parts of keeping track of connection state.
> > > > Naturally, we would also need to figure out a way to store the information
> > > > that the server would serve to make it useful, but to start with it's just a
> > > > neat real-world application of NIO.
No way!!! :-) Should read this last year when we were building
internal "cloud"!
It's all about file system (storage) ... great paper, thanks for
sharing!
~Jing
On Jan 3, 9:13 pm, Marshall Pierce <marsh...@software.mpierce.org>
wrote:
> Yeah, I think TCP is the way to go. There are some interesting things
> about reading from local storage but I get the feeling more people are
> interested in networking.
> To avoid schedule conflicts, the next class will now be on the *18th*.
> See you folks then.
> On Jan 3, 8:42 pm, Jing <sheeys...@gmail.com> wrote:
> > Marshall,
> > I assume you will teach/or touch a lot TCP/IP socket/streaming while
> > you teach NIO?
> > ~Jing
> > On Jan 3, 1:41 pm, Marshall Pierce <marsh...@software.mpierce.org>
> > wrote:
> > > I'm finally back in California -- had a lot of fun with airlines over
> > > the holidays. Nothing like being on vacation without your luggage.
> > > I'm trying to get everything together for a class on Jan 11th. Lot of
> > > work still to do on the NIO project -- see the aforementioned note
> > > about lack of luggage. :) I'll also have some coverage of Scala
> > > actors. I looked into Groovy as well during the break, but the
> > > language doesn't impress me. If there's interest, though, I could do a
> > > little bit on gpars.
> > > On Dec 14 2010, 2:17 am, Doug Chang <doug.ch...@hackerdojo.com> wrote:
> > > > Thank You Marshall: the delay is perfect. Feel free to delay to the second
> > > > week in Jan. Still haven't finished the fork-join homework.
> > > > Great idea on the NIO DNS server!!! I think this is my favorite class of all
> > > > time!
> > > > dc
> > > > On Mon, Dec 13, 2010 at 11:27 PM, Marshall Pierce <
> > > > marsh...@software.mpierce.org> wrote:
> > > > > The bad news: still no class on Tuesday -- I don't have enough prepared on
> > > > > the NIO project to take up a full class. Also, the following week I'll be
> > > > > out of town and I won't be back until the 29th, so that means the next time
> > > > > I'll be in town will be January 4th. I hate to not have class for that long,
> > > > > so if anyone has ideas on something they'd like to present in my absence,
> > > > > please propose them!
> > > > > The good news: I'm finally done with the interviewing/job hunt process as
> > > > > of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm
> > > > > stunned at how much time it took... I guess because it wasn't associated
> > > > > with being in an office, it never was really done for the day.
> > > > > Speaking of the NIO project, I've got a cool (and useful) project for us to
> > > > > work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with
> > > > > IPV6 for various reasons. I think it would be interesting to make a DNS
> > > > > server implementation that (1) works with IPV6 and (2) is as scalable as we
> > > > > can make it. Such a server would need to handle a crazy amount of
> > > > > requests/sec, but each request is just a UDP packet (no TCP connections!).
> > > > > Conveniently, this means we get to focus on scalability without worrying too
> > > > > much about the annoying parts of keeping track of connection state.
> > > > > Naturally, we would also need to figure out a way to store the information
> > > > > that the server would serve to make it useful, but to start with it's just a
> > > > > neat real-world application of NIO.
Well, if have a discussion on distributed file system, lean to
scalability solution, that will be even more interesting!
1 is easy, the road to N is tough, especially on sharing data
resource! :-)
Doug, how google implement its Meta-Data service/server across million
boxes? :-)
~Jing
On Jan 3, 9:13 pm, Marshall Pierce <marsh...@software.mpierce.org>
wrote:
> Yeah, I think TCP is the way to go. There are some interesting things
> about reading from local storage but I get the feeling more people are
> interested in networking.
> To avoid schedule conflicts, the next class will now be on the *18th*.
> See you folks then.
> On Jan 3, 8:42 pm, Jing <sheeys...@gmail.com> wrote:
> > Marshall,
> > I assume you will teach/or touch a lot TCP/IP socket/streaming while
> > you teach NIO?
> > ~Jing
> > On Jan 3, 1:41 pm, Marshall Pierce <marsh...@software.mpierce.org>
> > wrote:
> > > I'm finally back in California -- had a lot of fun with airlines over
> > > the holidays. Nothing like being on vacation without your luggage.
> > > I'm trying to get everything together for a class on Jan 11th. Lot of
> > > work still to do on the NIO project -- see the aforementioned note
> > > about lack of luggage. :) I'll also have some coverage of Scala
> > > actors. I looked into Groovy as well during the break, but the
> > > language doesn't impress me. If there's interest, though, I could do a
> > > little bit on gpars.
> > > On Dec 14 2010, 2:17 am, Doug Chang <doug.ch...@hackerdojo.com> wrote:
> > > > Thank You Marshall: the delay is perfect. Feel free to delay to the second
> > > > week in Jan. Still haven't finished the fork-join homework.
> > > > Great idea on the NIO DNS server!!! I think this is my favorite class of all
> > > > time!
> > > > dc
> > > > On Mon, Dec 13, 2010 at 11:27 PM, Marshall Pierce <
> > > > marsh...@software.mpierce.org> wrote:
> > > > > The bad news: still no class on Tuesday -- I don't have enough prepared on
> > > > > the NIO project to take up a full class. Also, the following week I'll be
> > > > > out of town and I won't be back until the 29th, so that means the next time
> > > > > I'll be in town will be January 4th. I hate to not have class for that long,
> > > > > so if anyone has ideas on something they'd like to present in my absence,
> > > > > please propose them!
> > > > > The good news: I'm finally done with the interviewing/job hunt process as
> > > > > of Tuesday, so now I'll have more time to prepare classes. Looking back, I'm
> > > > > stunned at how much time it took... I guess because it wasn't associated
> > > > > with being in an office, it never was really done for the day.
> > > > > Speaking of the NIO project, I've got a cool (and useful) project for us to
> > > > > work on. As it happens, the DNS RBL system (spamhaus, etc) doesn't work with
> > > > > IPV6 for various reasons. I think it would be interesting to make a DNS
> > > > > server implementation that (1) works with IPV6 and (2) is as scalable as we
> > > > > can make it. Such a server would need to handle a crazy amount of
> > > > > requests/sec, but each request is just a UDP packet (no TCP connections!).
> > > > > Conveniently, this means we get to focus on scalability without worrying too
> > > > > much about the annoying parts of keeping track of connection state.
> > > > > Naturally, we would also need to figure out a way to store the information
> > > > > that the server would serve to make it useful, but to start with it's just a
> > > > > neat real-world application of NIO.
