My thoughts about GNU WebPublish

[Gerhard Poul]

After seeing the announcement of GNU WebPublish I wanted to get more
infos about this new GNU project because I'm currently using a bunch
of makefiles + rsync to do what it does and more.

After reading the manual of GNU WebPublish (and I thank the authors
explicitly for writing a really good one) I think this is the most
un-GNU project I've seen published in a while by the GNU project.

These are the things that I don't really like about it:

- Uses a central database

  I don't think it's a bad thing to have a database, although I prefer
  text files. I just don't want to use a binary program to interface
  to a database if I could instead just edit some config files using
  emacs of vi.

  The bad thing with this database is that it is stored centrally in
  $HOME/.webpublish. Most often, if I have the chance to avoid programs
  who write into $HOME/.whatever, I do.

  IMHO it's much nicer to have it stored by-project in the same
  directories where the html pages are.

- Passwords seem to be saved in the database.

- To use a shell program to modify a file is not trivial

  You use the following command to use a shell program to modify a
  file before being uploaded:

    webpublish --add -F shell -a personal -e html -P /home/mydir/changebase

  Maybe this is just my personal preference but I think makefiles are
  much better suited for this purpose and also easier to use.

IMHO there are other programs that are more GNU-like out there like
sitecopy. (http://www.lyra.org/sitecopy/) Sitecopy even supports
WebDAV and is already included in Debian stable.

I still prefer rsync over ssh, but for people who don't have shell
access to their remote machines and have to use FTP or WebDAV one of
these tools is definitely the way to go.

These are just my collected thoughts about this tool. Maybe other
people like the design decisions that were made by the author/s of GNU
webPublish, but I don't.

yours,
  gpoul