llvm dev - Nov 2011 - [LLVMdev] Suggestions for LLVM Developer's Conference 2012

David,

Sorry I missed the Community Event Planning session.  There was this neat
session on Backend\Infrastructure competing with it.  :)

Whiteboards (with markers (and erasers?)).  
I was speaking with Dan G about the complexities of trying to get the
SelectionDAG to represent physical registers.  He mentioned during our
conversation that a whiteboard would help.  Ballroom Salon V would have been a
perfect place for a whiteboard room.  Surely others had conversations where if I
could just draw a box... with some arrows... this concept would totally make
more sense.  No DNE allowed ;)

Hacking Session
It was an absolute blast getting to meet many of the active contributors.  I
think it was a great meet & greet, and the room we were in was amply stocked
with refreshments.  Great job on getting that going.

But if we want to get some collaborative coding, must have a conference room
with tables, whiteboards, and wifi.  Also there wasn't much direction
towards the hacking, we were all coming from different backgrounds and
concentrations.  I think if there were some air of games/competition we'd
get more coding.  Perhaps something like make this C code run as fast as
possible, or as small as possible on x86 using only clang and llvm to compile. 
Patches welcome to make the code smaller/faster.   Another thought is to have a
C/C++ snippet which is laden with errors and or warnings, and enhance clang to
diagnose what is wrong.  Probably want to have some lottery-style organization
of teams, so that the community is strengthened with teams consisting of various
backgrounds.

On the other hand, I bet most of the attendees would probably go for boardgame
session.  Many boardgames, like Settlers of Catan, Bohnanza, Ticket to Ride,
Race to the Galaxy, etc etc exercise logic and strategy.  I think that would be
another way to keep conversations going.

Maybe others have thoughts, too.

Joe Abbey
Software Architect
Arxan Technologies, Inc.
1305 Cumberland Ave, Ste 215
West Lafayette, IN 47906
jabbey at arxan.com
www.arxan.com



-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111120/cbf59d61/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4350 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111120/cbf59d61/attachment.bin>

One idea for a hacking session would be a "performance analysis
workshop".  People could bring their apps, we could sample them track down
what part of the compiler would need to change and code it up (if time allowed).

Given the trade offs involved, it could be helpful to many folks, the trick is
to get the right people to show up.

-Chris

On Nov 20, 2011, at 2:10 PM, Joe Abbey <jabbey at arxan.com> wrote:
> David,
> 
> Sorry I missed the Community Event Planning session.  There was this neat
session on Backend\Infrastructure competing with it.  :)
> 
> Whiteboards (with markers (and erasers?)).  
> I was speaking with Dan G about the complexities of trying to get the
SelectionDAG to represent physical registers.  He mentioned during our
conversation that a whiteboard would help.  Ballroom Salon V would have been a
perfect place for a whiteboard room.  Surely others had conversations where if I
could just draw a box... with some arrows... this concept would totally make
more sense.  No DNE allowed ;)
> 
> Hacking Session
> It was an absolute blast getting to meet many of the active contributors. 
I think it was a great meet & greet, and the room we were in was amply
stocked with refreshments.  Great job on getting that going.
> 
> But if we want to get some collaborative coding, must have a conference
room with tables, whiteboards, and wifi.  Also there wasn't much direction
towards the hacking, we were all coming from different backgrounds and
concentrations.  I think if there were some air of games/competition we'd
get more coding.  Perhaps something like make this C code run as fast as
possible, or as small as possible on x86 using only clang and llvm to compile. 
Patches welcome to make the code smaller/faster.   Another thought is to have a
C/C++ snippet which is laden with errors and or warnings, and enhance clang to
diagnose what is wrong.  Probably want to have some lottery-style organization
of teams, so that the community is strengthened with teams consisting of various
backgrounds.
> 
> On the other hand, I bet most of the attendees would probably go for
boardgame session.  Many boardgames, like Settlers of Catan, Bohnanza, Ticket to
Ride, Race to the Galaxy, etc etc exercise logic and strategy.  I think that
would be another way to keep conversations going.
> 
> Maybe others have thoughts, too.
> 
> Joe Abbey
> Software Architect
> Arxan Technologies, Inc.
> 1305 Cumberland Ave, Ste 215
> West Lafayette, IN 47906
> jabbey at arxan.com
> www.arxan.com
> 
> 
> 
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111120/b38c3d61/attachment.html>

On 11/20/2011 4:33 PM, Chris Lattner wrote:
> One idea for a hacking session would be a "performance analysis 
> workshop".  People could bring their apps, we could sample them track 
> down what part of the compiler would need to change and code it up (if 
> time allowed).
>
> Given the trade offs involved, it could be helpful to many folks, the 
> trick is to get the right people to show up.
Chris, is there a formal feedback mechanism for the Developer's 
meeting?  One comment I'd like to make is that I think we need 
microphones both for the speakers at the conference and for people in 
the audience to use to ask questions at the end.  Every speaker had to 
repeat questions because audience members couldn't hear the questioner.

Of course, I think it speaks to LLVM's success that the meeting has such 
a problem: so many people attend that we have to get large rooms where 
hearing everyone is not possible.
:)

-- John T.
>
> -Chris
>
> On Nov 20, 2011, at 2:10 PM, Joe Abbey <jabbey at arxan.com 
> <mailto:jabbey at arxan.com>> wrote:
>
>> David,
>>
>> Sorry I missed the Community Event Planning session.  There was this 
>> neat session on Backend\Infrastructure competing with it.  :)
>>
>> *Whiteboards (with markers (and erasers?)). *
>> I was speaking with Dan G about the complexities of trying to get the 
>> SelectionDAG to represent physical registers.  He mentioned during 
>> our conversation that a whiteboard would help.  Ballroom Salon V 
>> would have been a perfect place for a whiteboard room.  Surely others 
>> had conversations where if I could just draw a box... with some 
>> arrows... this concept would totally make more sense.  No DNE allowed
;)
>>
>> *Hacking Session*
>> It was an absolute blast getting to meet many of the active 
>> contributors.  I think it was a great meet & greet, and the room we
>> were in was amply stocked with refreshments.  Great job on getting 
>> that going.
>>
>> But if we want to get some collaborative coding, must have a 
>> conference room with tables, whiteboards, and wifi.  Also there 
>> wasn't much direction towards the hacking, we were all coming from 
>> different backgrounds and concentrations.  I think if there were some 
>> air of games/competition we'd get more coding.  Perhaps something 
>> like make this C code run as fast as possible, or as small as 
>> possible on x86 using only clang and llvm to compile.  Patches 
>> welcome to make the code smaller/faster.   Another thought is to have 
>> a C/C++ snippet which is laden with errors and or warnings, and 
>> enhance clang to diagnose what is wrong.  Probably want to have some 
>> lottery-style organization of teams, so that the community is 
>> strengthened with teams consisting of various backgrounds.
>>
>> On the other hand, I bet most of the attendees would probably go for 
>> boardgame session.  Many boardgames, like Settlers of Catan, 
>> Bohnanza, Ticket to Ride, Race to the Galaxy, etc etc exercise logic 
>> and strategy.  I think that would be another way to keep 
>> conversations going.
>>
>> Maybe others have thoughts, too.
>>
>> Joe Abbey
>> Software Architect
>> Arxan Technologies, Inc.
>> 1305 Cumberland Ave, Ste 215
>> West Lafayette, IN 47906
>> jabbey at arxan.com <mailto:jabbey at arxan.com>
>> www.arxan.com <http://www.arxan.com>
>>
>>
>>
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu <mailto:LLVMdev at cs.uiuc.edu>
http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111120/92649d68/attachment.html>

On Mon, Nov 21, 2011 at 1:41 AM, James Molloy <james.molloy at arm.com>
wrote:
> Hi John,****
>
> ** **
>
> While this is a good idea, wouldn’t it logistically take longer to get the
> microphone to the person involved than for the presenter to repeat the
> question?
>
Have folks queue up in front of a fixed microphone?
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111121/fe4b19d1/attachment.html>

Possible, but wouldn’t it make the question session less ad-hoc and raise the
required activation energy for people to ask questions in the first place?

 

It would also make it harder to ask followup questions...

 

From: Chandler Carruth [mailto:chandlerc at google.com] 
Sent: 21 November 2011 10:04
To: James Molloy
Cc: John Criswell; llvmdev at cs.uiuc.edu
Subject: Re: [LLVMdev] Suggestions for LLVM Developer's Conference 2012

 

On Mon, Nov 21, 2011 at 1:41 AM, James Molloy <james.molloy at arm.com>
wrote:

Hi John,

 

While this is a good idea, wouldn’t it logistically take longer to get the
microphone to the person involved than for the presenter to repeat the question?

 

Have folks queue up in front of a fixed microphone?
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111121/e6e1d02b/attachment.html>

On 21.11.2011, at 10:43, James Molloy <james.molloy at arm.com> wrote:
> While this is a good idea, wouldn’t it logistically take longer to get the
microphone to the person involved than for the presenter to repeat the question?
Have enough microphones and volunteers who deliver them so the mic for
the next person arrives during the discussion of the previous
question? Not sure how well that model scales though.

- Ben

On 11/21/11 4:04 AM, Chandler Carruth wrote:
> On Mon, Nov 21, 2011 at 1:41 AM, James Molloy <james.molloy at arm.com 
> <mailto:james.molloy at arm.com>> wrote:
>
>     Hi John,
>
>     While this is a good idea, wouldn’t it logistically take longer to
>     get the microphone to the person involved than for the presenter
>     to repeat the question?
>
>
> Have folks queue up in front of a fixed microphone?
This was what I was actually thinking.  I've seen this done at academic 
conferences (e.g., Usenix Security), and it works pretty well.

It does have the downside that people in the middle of each row have a 
harder time getting to a mic, but since people usually know for which 
talks they'll have questions, it seems to work okay in practice.

Also note that it's probably hard for speakers to hear questions as well.

-- John T.

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111121/88d22cd6/attachment.html>

> -----Original Message-----
> From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at
cs.uiuc.edu]
> On Behalf Of Benjamin Kramer
> Sent: Monday, November 21, 2011 2:56 AM
> To: James Molloy
> Cc: llvmdev at cs.uiuc.edu
> Subject: Re: [LLVMdev] Suggestions for LLVM Developer's Conference 2012
> 
> On 21.11.2011, at 10:43, James Molloy <james.molloy at arm.com>
wrote:
> 
> > While this is a good idea, wouldn’t it logistically take longer to
> get the microphone to the person involved than for the presenter to
> repeat the question?
> 
> Have enough microphones and volunteers who deliver them so the mic for
> the next person arrives during the discussion of the previous
> question? Not sure how well that model scales though.
IME only one mic is needed (we always ran into problems with more than one as to
order of questions).

Greg