incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Emmanuel Lecharny <>
Subject Re: [DISCUSS] real-time communications
Date Wed, 24 Nov 2010 23:58:46 GMT
On 11/24/10 7:13 PM, Noel J. Bergman wrote:
>> The premise of this discussion is that running Apache projects are
>> *permitted* to engage in real-time communications, so long as they
>> take due care to avoid community problems of exclusion and closed
>> decision making.
> Did you see Greg's e-mail?
> "Just bring a summary of discussion points back to the list, along with any recommendations.
 The list can then sort through it and make decisions."
> Decisions are not made except on the mailing lists.  If it starts to seem that people
are being excluded from being an effective part of a decision process, curtail or modify the
back-channel communications.
>> We all want strong communities that are inclusive and open. We all recognize that
real-time communications pose risks to that.
> +1
> FWIW, ApacheDS and Geronimo had (possibly still have) *very* active IRC channels (logged)
where people talked in real-time, just as they might at a Hack-a-Thon.
Talking about ApacheDS, yes, we *do* use IRC actively. This is for us 
the main stream when we are working on some part of the server, plus a 
way to have a semi-F2F discussion about technical points. But this is 
like if you are working in an office in front of a collegue : most of 
the time, you don't chit-chat, you work. And when it comes to make a 
decision impacting the whole project, then the discussion is moved to 
the ML.

So why are we using IRC for ? Mainly for technical discussions. As an 
example, here is a short snippet :
elecharny: kayyagari: I'm wondering if we coudn't describe the tests 
using JSON
[6:38pm] kayyagari: hmm, am thinking of automatic pdu generation in 
tests for sequences that use components
[6:38pm] kayyagari: so only the lowest components need to be tested with 
hand crafted PDUs
[6:40pm] kayyagari: and cause lowest components definitely be small in size
[6:40pm] kayyagari: (atleast that is the case so far)
[6:45pm] elecharny: what is important is to test that PDU don't crash 
when there are some missing optional elements
[6:45pm] elecharny: the lower components have been quite well tested
[6:46pm] elecharny: the main problem with test generation is to write 
the PUD by hand,
[6:46pm] elecharny: and to compute the lengths
[6:46pm] elecharny: one thing we can do is to use the DERxxx classes we 
have in the project
[6:47pm] elecharny: you can create a PDU using them, with no issues like 
creating the lengths
[6:47pm] kayyagari: ahah
[6:47pm] kayyagari: am trying to computelength() and then encode them to 
a temporary buffer and later push it to the main buf
[6:47pm] elecharny: you can do that
[6:47pm] elecharny: otherwise, you can also do :
[6:48pm] elecharny: seq = new DERSequence();
[6:49pm] elecharny: seq.add( DerInteger.valueOf( 5 ) );
[6:49pm] elecharny: etc...
[6:49pm] kayyagari: aha, will try it, this is helpful
[6:49pm] elecharny: and at the end do a seq.encode( outputStream );
[6:49pm] elecharny: you'll get the PDU
[6:49pm] kayyagari: cool
[6:50pm] kayyagari: this is the class present in shared-ldap?
[6:50pm] elecharny: but you won't be able to test cornercase like 
Integer with no values
[6:50pm] elecharny: yes
[6:50pm] kayyagari: ok
[6:50pm] elecharny: in asn1.der
[6:50pm] kayyagari: yeah
[6:51pm] elecharny: ok, I have to run, some friends have flown from 
Boston to paris

Nothing fancy, just raw technical convos. But very useful when you have 
a huge code base and you want to share some informations IRT.

We tried to organize Skype sessions 4 years ago, but we never succeeded 
to get them lasting more than 2 or 3 weeks. The reasons are simple :
- with people in South Korea, France and Florida, the TZ pb is almost 
impossible to solve
- as the schedule was very tight, being 15 minutes late just blows the 
- surprisingly, no more than 15% of the world population is speaking a 
fluent english. That does not help when on skype
- You always have to find someone to report what has been said, and 
trust me, a lot of input is lost in the process

We decided that it was a waste of time. I still think it's a waste of 
time, and also a perfect way to split the community in 2 groups :
- those who participate
- those who are left behind and get frustrated.

IRC solve some of those issues : it's easier to understand what a poor 
english speaker like me is typing, it's easier to get a report and move 
it back to the ML, and also it's easier to get the discussion on track, 
without the one who speak louder talking most of the time.

Emmanuel L├ęcharny

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message