httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ben Hyde <bh...@pobox.com>
Subject Re: Acknowledgements
Date Wed, 16 Dec 1998 13:54:37 GMT
On the topic of people stuffing their nom_de_guerre into the noble
sources.  I've had this argument before so here is my rant on
the topic.  You can read the first paragraph and skip the rest.

If you are clear that names on the code are an honoraria and not a
source control lock then things are fine.  You can then encourage
people to sign their work at the routine, and file level - to add
their name when they make significant revisions.  This gives a small
gift to their reputation and increases the quality of the effort all
around.

As labor it has always seems rude, cheap, and greedy that management
didn't want my name confusing their ownership of the code.  That I
would spend months getting some hair ball like buff.c to work in a
reasonable number of cases and the boss would then announce - "I'll
take that, you can go now."  There is one large system I built that
had a policy against this - hundreds of thousands of lines of code and
people there almost don't know who I am!  That hurts.

As a manager I've always encouraged it.  Why wouldn't I want people
to: develop a sense of ownership, pride in the code, and take on some
risk to their reputation when things don't work out.  As a capitalist
I consider it a way to increase lock-in between the developer and the
code so it becomes harder for him to switch to some other bondoggle.

As a user of code I'm always very glad to see an author's name since
it gives me - more than you know - a sense of what to expect.  No
amount of coding conventions can override the deep style of each
programmer.  When I go to work on code has the certain style I want
to stay close to that style as I change it and any hints are a help.
Then of course there is always the hope that i can ask the subtle
question of the author - such as "Is it a good witch or a bad witch
that ap_bnonblock is called only once?"

The only risk I consider credible here is that other developers will
perceive the names on the code as a "no trespassing" sign.  Of course
all working code has an implicit "authorized service personnel only"
sign on it.  The names are not the same as "authorized," no author
want that.  Only the youngest authors working on their first program
would want that.  The most wonderful thing in the world is getting
somebody else to pick up where you left off.

The problem, a central problem, with all mature code is that
developers have a natural reluctance to cross over into it due to:
fear of embarrassment, fear of missing authorization, fear of trespass
on another's turf, fear of having to suffer long lectures on
esoterica, etc., etc.  Grid lock often results.  This really doesn't
have much to do with names on the code.

Yes, names on the code can create an excuse not to dive in and take a
wack at things.  But they are not the principle cause of this problem,
the principle cause is that bundle of fears.  In my opinion they are a
useful tool, a reward if you will, in overcoming that fear.

  - ben

Mime
View raw message