#12 jar files incorrectly treated as ascii

[Aslak Hellesřy]

When adding/committing a new jar file to cvs, cvs-
syncmail incorrectly sends a message like this:
------------
Update of /cvsroot/xdoclet/xdoclet2/lib/rngconv/jars
In directory sc8-pr-
cvs1:/tmp/cvsserv14703/lib/rngconv/jars

Added Files:
rngconv-20030225.jar
Log Message:
Initial support for XML Schema in XGG (using Sun's
rngconv)

--- NEW FILE: rngconv-20030225.jar ---
PK
------------
This leads people to think the file was incorrectly added
as ascii, while it's actually syncmail that incorrectly
assumes it's ascii.

(jar files are in zip format).

[Fred L. Drake, Jr.]

Logged In: YES
user_id=3066

This one's my fault to begin with, so I'm assigning it to
myself.
There are a couple of possible approaches to handling this:

1. Use the "rlog -f comma-v-file" command to get the log
header information from the RCS backend, looking for the
"keyword substitution" field. Pros: sticks with something
reasonable close to a "public API" for the system; we don't
have to reach into the actual file. Also, I've written code
to do this before. Con: We need to start up another
process for each syncmail run to get the information.

2. Parse the information directly from the ,v file. Pros:
It's faster; no additional processes to manage. I know
where I can easily get some code to do this from a public
source. Con: We have to deal with the format of ,v files
directly. Seriously, the ,v format just hasn't changed
recently, and isn't likely to do so.

[Fred L. Drake, Jr.]

Logged In: YES
user_id=3066

I've indicated elsewhere that I'll be fixing this to use
"rlog -h" (the -f was a typo) to fix this.