RFR: 8026062 : (s) webrev.ksh: fix bug title web scraping, remove teamware, sac, "open bug" and wxfile support

John Rose john.r.rose at oracle.com
Thu Oct 10 02:45:35 UTC 2013


On Oct 9, 2013, at 3:54 PM, Mike Duigou <mike.duigou at oracle.com> wrote:

> Hello all;
> 
> This changeset revisits webrev's integration with bugs.openjdk.java.net to correct problems with the scraping of bug titles from jbs html. I also eliminated the now obsolete "-O" option since bugs are now all visible on the preferred bugs.openjdk.java.net system. 
> 
> While I was at it I finally eliminated the obsolete teamware, sac and wxfile support with the hopes that cruft removal may lower the bar to others contributing to webrev (yeah, it's not glamourous but we still need it).
> 
> http://cr.openjdk.java.net/~mduigou/JDK-8026062/0

Great deletions!

Here's my "private reserve" version of webrev's html_quote, some of which you may want to use:

 #
 # Make a piece of source code safe for display in an HTML <pre> block.
+# If the line appears to start with an HTML tag, assume it is markup and don't quote on it.
 #
 html_quote()
 {
-	sed -e "s/&/\&/g" -e "s/</\</g" -e "s/>/\>/g" "$@" | expand
+	sed -e '/^<[a-zA-Z][^<]*[a-zA-Z_0-9\/"'\'']>/{p;d;}' \
+	    -e '/^&#*[a-zA-Z_0-9]*;/{p;d;}' \
+	    -e "s/&/\&/g" -e "s/</\</g" -e "s/>/\>/g" \
+	    "$@" \
+	| expand
 }
 
A few points about it:

If a line appears to already have HTML in it (heuristically!), it goes completely unquoted.  (This makes it easy to add markup to "include" files.)
The sed code for & handles consecutive rewrites, as "&&" or "<&>", which yours does not appear to do.
My Mac's version of sed does not appear to rewrite "&" to "&" using your code.

— John

P.S.  Like most folks, I have other tweaks in my webrev, but they would be out of scope for this change.

Here is the way I use webrev (in the setting of a workflow script):
  webrev-from-openjdk $force_mercurial -u $WEBREV_USER -o $WEBREV_DIR -N -r -2

The "off by one" revision is very important.  It selects the parent of the revision I want to review.  I use mq to manage changes all the way through review, so "hg outgoing" is irrelevant to me.  Instead, I "hg qgoto" the change I want to review, to make it the tip, and then review against -2.  This seems to me far safer and more flexible (and with fewer Merge changesets) than the alternatives.

I also have a push-button workflow script which can generate the next (00, 01, 02, ...) minor webrev version for a given bug and push it to cr.ojn.  It seems like that should be easy to do with an OpenJDK savvy webrev.




More information about the build-dev mailing list