Any YARV pre-questions in RubyConf2004?

S

SASADA Koichi

Hi,

I will show presentation about "YARV - Yet Another RubyVM"
in RubyConf2004.

But I have no enough topic to talk about that for 40 mins full.

Does anyone have any question about YARV or something I can
answer? If you gime me that, I'll prepare to answer that.

(And because I can't listen English well, answering direct
is too hard for me :X)


I look forward to seeing you > RubyConf2004 attendances

Regards,
 
J

James Britt

SASADA said:
Hi,

I will show presentation about "YARV - Yet Another RubyVM"
in RubyConf2004.

But I have no enough topic to talk about that for 40 mins full.

Unless you've given this presentation before, time estimates can be hard
to get right. You may end up running longer than you think.

And if you don't use a full 40 minutes, it may balance out if the other
speakers are running longer than their alloted time.
Does anyone have any question about YARV or something I can
answer? If you gime me that, I'll prepare to answer that.

It seems to me that just the general idea of a VM would take a fair
amount of time to explain, particularly if he VM is for a dynamic
language. Anyway, here are some questions:

Why bother?

Where do you begin when designing a VM?

What are the options?

Are there some common VM architectures?

Why not just target the Java or .Net (or compatible) VMs?

How do you handle dynamic class changes (for example, adding a new
method to String at run time)?

What are the really hard parts?

What surprised you?



James
 
E

Eivind Eklund

Hi,

I will show presentation about "YARV - Yet Another RubyVM"
in RubyConf2004.

But I have no enough topic to talk about that for 40 mins full.

Does anyone have any question about YARV or something I can
answer? If you gime me that, I'll prepare to answer that.

I won't be at RubyConf (unfortunately), but I've still got one
question: Does the structure of YARVM allow a gradual transition from
the standard Ruby node structure to JITed code, or is it a Big Bang
approach?

Eivind.
 
D

David A. Black

Hi --

Unless you've given this presentation before, time estimates can be hard
to get right. You may end up running longer than you think.

And if you don't use a full 40 minutes, it may balance out if the other
speakers are running longer than their alloted time.

Perish the thought! :)


David
 
J

Jamis Buck

David said:
Hi --




Perish the thought! :)

:) I'm actually kind of worried. I did a preview demonstration of my
presentation to my local RUG, and it went for almost exactly an hour...

Have your tomatoes handy, if I start going overtime. :)
 
A

Alexander Kellett

But I have no enough topic to talk about that for 40 mins full.

i've got a large number of questions that you may find useful :)
Does anyone have any question about YARV or something I can
answer? If you gime me that, I'll prepare to answer that.

here's the beginning. i'll think of somemore to ask in the
coming days :)

will the standard library be written in ruby or c?

what is the time frame you are aiming to keep to?
any idea when you'll see results?

what ruby code can yarv currently execute?

do you plan to perform code/bytecode optimisations
during execution or only at compile time? if so,
what sort of optimisation do you think you will
be able to offer?
(And because I can't listen English well, answering direct
is too hard for me :X)

:/ i should practice my japanese (i know none :/) then as even
native english speakers sometimes have difficulty understand me :(
I look forward to seeing you > RubyConf2004 attendances

unfortunately won't be able to make it :(
i hope that next year we can meet!

Alex
 
Y

Yukihiro Matsumoto

Hi,

In message "Re: Any YARV pre-questions in RubyConf2004?"

|> But I have no enough topic to talk about that for 40 mins full.
|
|Unless you've given this presentation before, time estimates can be hard
|to get right. You may end up running longer than you think.

Don't underestimate the language barrier. ;-)
I know exactly what he feels.

Be nice to him, please, conference attenders.

matz.
 
C

Chad Fowler

Hi,

In message "Re: Any YARV pre-questions in RubyConf2004?"

|> But I have no enough topic to talk about that for 40 mins full.
|
|Unless you've given this presentation before, time estimates can be hard
|to get right. You may end up running longer than you think.

Don't underestimate the language barrier. ;-)
I know exactly what he feels.

Be nice to him, please, conference attenders.

Speaking for everyone I think, we'll be honored to have him as our
guest at RubyConf. We value all of the attendees, but there's
something special about "original" Rubyists coming _all the way_ from
Japan.

We'll miss you this year, Matz.

Chad
 
J

James Britt

Yukihiro said:
Hi,

In message "Re: Any YARV pre-questions in RubyC onf2004?"

|> But I have no enough topic to talk about that for 40 mins full.
|
|Unless you've given this presentation before, time estimates can be hard
|to get right. You may end up running longer than you think.

Don't underestimate the language barrier. ;-)
I know exactly what he feels.

Be nice to him, please, conference attenders.

Well, OK, since you asked so nicely.

:)

Having people pose questions here in advance should be a big help, so if
folks can think of anything, please speak up.

I saw Dan Sugalski give a talk in Parrot at local Perl users group, and
the topic of VMs can get quite deep.

Selection of architecture, byte code set, the compilation process; stuff
I used to know more about, once upon a college time, but has faded from
memory

It's going to be a challenge as it is, with the language barrier, so any
suggested topic or question should be quite useful. It's often hard to
know in advance just how familiar your audience is with your topic, and
it's tough to walk the line between stating the obvious and baffling the
crowd.

As for my questions, while I think a VM talk should probably explain why
one would bother, I also think we all have god ideas as to why this is a
Good Thing. And I'd be curious to know why the design does or doesn't
target an existing or expected VM/byte-code-set. But really I want to
understand the thinking process that lets one take an interpreted
language and move it to a VM.

James
 
Y

Yukihiro Matsumoto

Hi,

In message "Re: Any YARV pre-questions in RubyConf2004?"

|We'll miss you this year, Matz.

I miss you all too. But I shouldn't miss my wife and baby this year.
I hope I can see you guys next year.

matz.
 
J

James Britt

Yukihiro said:
Hi,

In message "Re: Any YARV pre-questions in RubyConf2004?"

|We'll miss you this year, Matz.

I miss you all too. But I shouldn't miss my wife and baby this year.
I hope I can see you guys next year.


In Toronto!

Right, folks?

:)


James
 
A

Austin Ziegler

Everything's west of somewhere.

I'll push for 2006 :) I might be able to get some corporate
sponsorship on board for then, too :) (I think I might be close to
convincing work to use Ruby as our build/test scripting language. This
may ultimately involve porting to AS/400 and NetWare... :) )

-austin
 
S

SASADA Koichi

James Britt <[email protected]>
Tue, 28 Sep 2004 06:29:08 +0900 / Re: Any YARV pre-questions in RubyConf2004?

Hi,
Thank you for many replies! > ruby-talk rubyist

Unless you've given this presentation before, time estimates can be hard
to get right. You may end up running longer than you think.

I have done sometimes, but only once in English (that first speach is very poor :X)
It seems to me that just the general idea of a VM would take a fair
amount of time to explain, particularly if he VM is for a dynamic
language. Anyway, here are some questions:
Why bother?
Where do you begin when designing a VM?
What are the options?
Are there some common VM architectures?
Why not just target the Java or .Net (or compatible) VMs?
How do you handle dynamic class changes (for example, adding a new
method to String at run time)?
What are the really hard parts?
What surprised you?

I'll talk about these.

Thanks,
 
S

SASADA Koichi

Eivind Eklund <[email protected]>
Tue, 28 Sep 2004 06:41:05 +0900 / Re: Any YARV pre-questions in RubyConf2004?

Hi,
I won't be at RubyConf (unfortunately), but I've still got one
question: Does the structure of YARVM allow a gradual transition from
the standard Ruby node structure to JITed code, or is it a Big Bang
approach?

I use ruby interpreters parser (in C interface) and compile this AST to
my bytecodes (in C).

JIT compiler isn't available now. (only planning)
AOT compiler seems more easier to implement. (bytecode -> C code)
 
G

gabriele renzi

Austin Ziegler ha scritto:
use Ruby as our build/test scripting language. This
may ultimately involve porting to AS/400 and NetWare... :) )

join the damagecontrol army :)
 
G

gabriele renzi

SASADA Koichi ha scritto:
Hi,

I will show presentation about "YARV - Yet Another RubyVM"
in RubyConf2004.

But I have no enough topic to talk about that for 40 mins full.

Does anyone have any question about YARV or something I can
answer? If you gime me that, I'll prepare to answer that.

thank you for yarv and for asking our feedback.. emh.. feedforward maybe..

I'd ask:
- what is yarv aiming to optimize? method calls, proc calls, yielding,
numerical stuff and how much (say, 20 times faster callcc)
- is yarv somewhat similar to something else? (so that I can look up
documentation of concepts even if you don't explain them yourself widely)
- is the yarv design open to further enhancements? I don't even really
know what this mean, to be fair :) but I think something like "well, we
can explore this kind of optimization in the future"
- can yarv support the existing 'infrastructure' stuff like
set_trace_func, debug.rb and such ?
- is yarv general enough to support something else? (remembering some
irc chatting about scheme :)
 
G

gabriele renzi

SASADA Koichi ha scritto:

AOT compiler seems more easier to implement. (bytecode -> C code)

then another question:
- how fast this would be?
IIRC there was a python2c compiler, but it meet the same barriers that
the python VM hits, like loosing lot of time in method lookup. It has
the advantage of obfuscate the code, for those who want it, but little
performance gain.
I remeber this from reading the StarKiller paper (python->c++
compiler+type_inferencer)
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

No members online now.

Forum statistics

Threads
473,744
Messages
2,569,482
Members
44,901
Latest member
Noble71S45

Latest Threads

Top