A note were in Ben has post Vietnam flashback...
Justin Erenkrantz wrote:
Leo Simons <[EMAIL PROTECTED]> wrote:
NNTP makes more sense than SMTP for group discussions.
No, it doesn't necessarily.
I was in a room yesterday with a lot of people who have been spending way too much time attempting to 'understand open source'. I am always greatly amused by these gathering because you have a group of extremely smart people attempting to tease apart the mystery of what we do. That's fascinating to listen too. But they are always attempting to do so with out getting the hands dirty. At one point during the day they spent a good two hours discussing one thread in the Linux kernel mailing list. They looked at the first four messages and then were given a summary of the other 175 messages. I was the only person in the room that had the least clue what the topic of the thread was.
What I often say in these gathering is that Open Source is both a very old process, i.e. people cooperating to solve problems, and a very new one, i.e. manufacturing of a document intermediated by computers. We are a long way from having searched the space of all possible schemes for doing it effectively. That even the simplest things, like what affordances on the tools might help are just beginning to get some experimentation.
For example would open source projects work better if CVS was more enthusiastic about forking? Would a cooperative drawing tools help? Is there a right size for a repository, or a community, or an appropriate way to manage the boundaries around these? Nobody knows, but we have lots of intuitions that may or may not just be cargo cult experiences.
I have a lot of nostalgic affection for using NNTP rather than SMTP for this stuff. The first project I ever worked on that was effectively intermediated using computers and networks used NNTP. 1980-85 I worked on a very complex compiler project and we used NNTP to organize the narrative around the work. For example we used individual threads for each bug. We never discarded any messages. The entire message store was in flat files on the file system so we could use unix tools to search for lost bits of info - like "did heather say something about stack chunks a few weeks back". I vividly recall noticing about half way thru that project that we all stopped talking about the work in the halls, that we only talked about life in the halls. That struck some of the management as very odd.
We had stumbled on the discovery that by normalizing the work so it all happened in one place intermediated by the machine we were more thoughtful, more efficient, more convivial, effective due to the orthogonal skills, able to tap into bits of serendipity, able to work with people who could be prickly in person, etc. etc. etc.
This was also the first time I saw automated builds and testing used to help developers avoid embarrassment. The first time I saw builds and testing triggered by commits so we could shorten the interval between break and fix - and hence increase the chance that the guy that broke it could remember what he thought he was doing. It was also the first time I got the commits posted to netnews so many eyes could casually proof read them.
One thing that would happen in that context was that NNTP threads would often lie idle for months only to suddenly come back to life. That is one reason it could be used for working on individual bugs/enhancements/ports/etc. The threading seemed more robust then it it ever seems to in [EMAIL PROTECTED] I assume that was because we were all using the exact same client software then.
These days I have a lot of trouble seeing the difference between a good news setup and a good mail setup. But both need archive, threading, search etc. etc. I run all the [email protected] email into my in-house news server, for example.
Of course the major user list for httpd is netnews based.
- ben
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
