Docvert 3.4

[holloway]

Hi folks (I hope this is the right place to say this -- I searched for a contact address on the lemon8 site but couldn't find anything),

The Lemon8 FAQ says,

Q: what are the "substantial installation requirements" you describe?
A: unlike the rest of the PKP suite, Lemon8-XML doesn't have an easy installer interface; as well, the Java components (notably Apache FOP) require some admin experience to set up correctly. In addition, to convert documents from MS-Word or similar to OpenOffice ODT automatically, Docvert must be installed, which is fairly complicated to install and configure.

(emphasis mine)

Well I've taken that comment to heart and the new release of Docvert is much much (much) easier to install. Basically you just need to install some python libraries and click a button on the admin page now.

More generally, if there's anything that Docvert should be doing that it isn't then let me know and I'll see about improving it.

Cheers!

[codonnell]

Oh, now I am stoked, knowing that L8X has gotten Matthew's attention.

[mj]

Hi Matthew,

I'll definitely second Casey's comment in that it's a pleasure to have you on the L8X forum. I should also begin by clarifying that the wording in the FAQ isn't intended as a criticism of Docvert -- in fact, quite the opposite: much of the easy-to-use aspect of Lemon8-XML comes from building upon the incredible work you've done with Docvert.

As I'm sure you're familiar, the difficult aspect relates to the overall process of getting OpenOffice installed and configured on a server -- in many cases, remotely, and often headless without X11 available. You've certainly made a lot of it easier with a number of features in Docvert (eg. testing macro instantiation, providing suexec/sudo scripts, xvfb integration) and that has certainly lowered the barrier for many people. Unfortunately for others, much of our user base is extremely non-technical, and working in very restrictive environments; often shell access is a luxury, and installing libraries is impossible. There will also always be some cases where people will need to do run-time OO instantiation instead of using UNO (ie. with limited, shared server resources), and infinite permutations of platform issues.

That being said, I remain committed to Docvert as the best approach out there for making this kind of served-based document conversion possible. Your excellent documentation and tireless improvement of the software make it a natural partner tool for L8X, and if you've made it easier with the PyUNO approach, so much the better. We're releasing the L8X public beta this week -- the first time we've made the source code available (and, mea culpa, long overdue) -- so, we'll be sure to gather feedback on Docvert-L8X integration, as well as any issues people may have with installing Docvert for themselves. If there's anything we can identify that will make L8X or Docvert easier to use, let's be sure to keep in touch.

On this last note, do feel free to email me directly any time (my address is available at http://pkp.sfu.ca/people). Once again, great to hear from you, and I'm looking forward to trying out the new Docvert 3.4 release.

Best regards,
MJ (L8X Lead Developer)

[holloway]

@codonnell,

Yes, mark down yesterday in your calendar -- the day when a programmer dude from from New Zealand posted on this forum

@mj

Oh I consider that FAQ to be accurate. It is a pain to set it up. The new Docvert 3.4 way of doing it is considerably easier (no need to setup openoffice or trusted-macros, just install python-uno and throw some lines in sudoers) but it's still not as easy as it could be. We're looking at getting Docvert into Debian, and that should ease installers.

I was actually planning on removing the xvfb+OpenOffice.org option for 3.5... are you suggesting that it should be retained? If so, why? I agree with memory usage arguments (leaving OOo running in the background takes about 60MB for me) but in all other cases daemonizing OOo is much nicer. Perhaps I could unload the daemon if it's not being used for a while -- would that be ok?

We're releasing the L8X public beta this week -- the first time we've made the source code available (and, mea culpa, long overdue) -- so, we'll be sure to gather feedback on Docvert-L8X integration, as well as any issues people may have with installing Docvert for themselves. If there's anything we can identify that will make L8X or Docvert easier to use, let's be sure to keep in touch.

Sounds like a plan

[mj]

Hi Matthew,

We've actually submitted OJS for inclusion into Ubuntu, so a wonderful ideal would be for someone in the future to be able to do, eg.

Code: Select all

sudo apt-get install docvert lemon8

And be done with it. As for the 3.4/3.5 Python-UNO approach, you're in a much better position to know what Docvert users want, and what's easier for them to deploy. So, I'll happily defer to your judgement. The memory argument is a tricky one, since OOo is in one sense "heavy" (both in terms of memory and CPU) while it's running, but with the advent of virtual machines, and certainly compared to some larger server applications (eg. Java-based web applications), a daemonized OOo in 60MB isn't so bad at all.

Particularly if moving to Python-UNO obviates the need for messing around with X11 (including xvfb), then I'd say that's probably the way to go, with the rationale being that users who want to use Docvert should basically be setting up a dedicated "document conversion" server (or VM) as the trade-off for it being easier to set up.

Cheers,
MJ

[mj]

Hi all,

I wanted to share a recent notice posted by Matthew to the Docvert mailing list that might be useful to a number of people:

We're currently testing a Debian-based repo for Docvert so if you want to help out then try this and note down any errors or warnings you get or better ways of explaining things
WARNING: backup your existing install before doing this.

FOR UBUNTU

1) add this to your APT sources, /etc/apt/sources.list

Code: Select all

deb http://docvert.org/debian etch main

2) Run these commands

Code: Select all

apt-get install gnupg
gpg --keyserver subkeys.pgp.net --recv-key AAB0E0D1F4B68080
gpg -a --export AAB0E0D1F4B68080 | apt-key add -
apt-get update
apt-get install openoffice.org-headless docvert docvert-openoffice.org

3) Enable docvert web interface by uncommenting /etc/apache2/conf.d/docvert

4) Restart apache,

Code: Select all

apache2ctl configtest
apache2ctl graceful

FOR DEBIAN

1) add this to your APT sources, /etc/apt/sources.list

Code: Select all

deb http://mirror.linux.org.au/backports.org/ etch-backports main
deb http://www.backports.org/debian etch-backports main
deb http://docvert.org/debian etch main

2) Run these commands

Code: Select all

apt-get install gnupg
gpg --keyserver subkeys.pgp.net --recv-key AAB0E0D1F4B68080
gpg -a --export AAB0E0D1F4B68080 | apt-key add -
apt-get update
apt-get install openoffice.org-headless docvert docvert-openoffice.org

3) Enable docvert web interface by uncommenting /etc/apache2/conf.d/docvert

4) Restart apache,

Code: Select all

apache2ctl configtest
apache2ctl graceful

Hope this helps some folks get Docvert up and running more easily. Great work Matthew!

MJ