| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10
Comments:
<0> lol
<0> poor online retailer.
<0> should have never messed with Zet
<1> no
<1> it gonna cost them $78.544!
<0> hahha
<0> woahh
<1> thats what im gonna make them pay!
<0> 78k
<0> Zet,
<0> if you get that money
<0> can you buy me a PowerMac G4?
<1> lol :D
<0> ;) pleeeeaseeee?
<0> so I can make a PPC Port?
<1> hehe ok :P
<0> :D
<0> Thank you Zet!
<0> :D
<0> lol
<2> greets
<0> hey eieio
<0> Yknow how Intel mails you their Manuals?
<0> I wonder if Motorola or whoever does PowerPC now, does that
<2> hmm, i don't know i just use PDFs
<0> I dont have a powerpc so... I dont know how or where to start.
<0> :(
<0> I mean, if I developed soley on Qemu
<0> there is a high chance it would do didly sweet **** all on a real PPC machine
<2> hmm, maybe someday i'll release Monarc and then you can use that ;-)
<0> heh
<0> for what?
<2> http://www.coreytabaka.com/index.php?r_s=541&r_p=543&flog_id=13
<0> thats quite impressive, eieio
<2> ty
<0> id show you mine, but... mine is nothing impressive yet.
<0> atleast not from a graphical user standpoint.
<0> Might go and check the prices of old Mac computers.
<2> is your code available?
<0> well.. kinda
<0> its a mixed bag.
<0> the code is entirely my work
<0> but, I kinda protect it
<0> just incase its something cool
<0> lol
<2> ah, i understand your point of view
<0> I can tell you about Zenobit if youd like though, Im kinda proud of it.
<2> sure
<0> the overall design has been in the works for almost a year, but the sourcebase that is most recent has only been in developement since Oct/November last year
<0> WEll, Its a Microkernel designed from scratch to be fast, minimal, while being reliable, secure and modern.
<0> Its got a lot of things I consider cool.
<0> but, Im a perfectionist so I compulsively rewrite a lot of things once they lose their 'this is a cool thing' novelty.
<0> I just got the core IPC subsystem working tonight :D
<2> coo
<0> basically, my take on Messaging.
<0> Im designin gthe system as something I would want to use
<0> which si why I spend so much time rewritin gthings :P
<0> I Write it to be... fun to use as both a programmer, and eventually as a user :)
<2> that's generally a good approach
<0> Well , I guess its sad - but I see it like my baby :P
<2> how did you implement the IPC?
<0> well, what specifically do you want to know?
<0> I Went outside
<0> and watched how mailman mailed stuff
<0> and drew abstractions from that.
<0> :)
<0> Basically, I have a 4KiB mailbox ***osciated with every process, this can be easily expanded should I decide later.
<0> the idea is, the mailbox doesnt hold mail, as such
<0> it holds envelopes
<0> that the program checks
<0> the Mailbox is a FIFO stack. The envelope itself, has data about where the message came from, and where it really is, and its size.
<2> ah, ok i see
<0> when the program recieves the envelope, it is erased from the mailbox.
<0> however,
<0> the envelope is kinda cached in the program reading it
<0> so, they can get hte package :D
<0> thus, the data
<0> when they recieve the actual data, it too is removed.
<2> ok, so then there is a zero copy mechanism in the mix?
<0> basically how I Thought messaging would be cool.
<0> well, there is copying
<0> but there are limits I have placed on the Messaging system.
<0> Zenobit supports different kidns of IPC, for different situations.
<0> :P its meant to be superflexible.
<2> oh, ok
<0> how large the maximum message can be, is variable, I can set it whenever.
<0> but, the default at this time is 4Kb
<0> <4Kb, message is okay.
<0> >4kb, better options
<0> like, shared memory and such
<2> right right
<0> sound okay?
<0> If the Mailbox is full, there is a certain queue system in place.
<2> yeah, seems pretty reasonable
<0> when you recieve an envelope, anything that maybe queued, is put into the newly unful mailbox.
<0> Its a binary search tree indexed on process address, with a FIFO stack of envelopes.
<0> if a process has nothing queued anymore, its node is removed from the tree.
<2> what happens when you check an empty mailbox? are there blocking and nonblocking means to check one?
<0> at the moment, theres no need to block
<0> more layers will be built on the Messaging system
<0> if you check an empty mailbox, you just get told via the envelope, that you have no messages.
<0> but it wouldnt be hard to make it block.
<0> The next layer ontop of this, will be a Callback system,.
<0> Zenobit is designed in a way, that interrupts can be handled entirely in userspace, and such
<0> messages can be done just like interrupts.
<2> essentially allowing a process to recieve a notification of a message?
<0> yup, or a asynchronous way to handle them
<0> but thats a few weeks away.
<0> There are commands though, to yield thread timeslice though
<0> so the programmer can block his own program if he wants
<0> if he doesnt, itll get preempted anyway
<0> My ideology behind this is basically saying, Here developere, you have all of this freedom.
<0> at the same time, I have a lot of things in place to secure the system and make it as isolated as possible.
<0> Example of this is the Kernel itself, cant even use a Physical reference of memory.
<0> But, im proud of it, even prouder that is works quickly :)
<2> cool
<0> thanks
<2> so the messaging is the primary means of communication between processes?
<0> Well, I dont intend to use anymore of them more than the other
<0> I kinda expect to use them together
<0> Messaging, Ports, Shared Memory.
<0> and use them at the right times.
<0> IF you have to shift a great deal of data from one process to the other, then you should use shared memory.
<0> if you are sending a lot of little messages, all the time, ports.
<2> ah, do you have ports implemented yet then?
<0> if you are sending decent sized messages, use Messaging.
<0> Nope, but thats the next thing I implement.
<0> Kinda in the process of designing my implementation of them
<0> I have Semaphores too, so thats handy.
<0> My beef with the project at the moment is though, that its .. getting kinda big, atleast for me.
<0> entire OS Boot package, is tiny, about 47kb
<0> but the sourcebase, its quite large
<2> how many lines of code?
<0> to be honest, I cant give you an exact number.
<0> but two days ago, it was 10480
<0> and the Messaging system so far is about 600
<0> so, add that to it
<0> + about 800 lines
<2> ah, ok not bad
<0> its probably the largest thing ive ever written
<0> I just hope its worth it
<2> it'll be a good experience, no doubt about it
<0> Im kinda.. obsessive with it.
<0> I Dream about code :P
<0> and components
<0> I need for it, to not be a peice of **** ;P
<0> on the brightside, im learning a great deal :)
<2> Monarc is 91,460 lines at the moment
<0> about design, research and such
<0> woah
<0> how do you manage it all?
Return to
#osdev
or
Go to some related
logs:
#gentoo
t
mssql replace Msg 8114
#ubuntu
GPixPod ebuild
Can't locate somemodule.pm
openldap custom overlay
Fatal error: mysql.user table is damaged or in unsupported 3.20 format
ssh session localization linux
#python
|
|