Traditional data flow

  1. Data sent to CVS or/and submit list
  2. Data copied up to the ibiblio server
  3. Data published using Greg ldp_mk script. Entire toolchain is present on the ibiblio server. Script creates all variants: HTML, HTML single file, PDF, Postscript, text, pluckerDB).
  4. Data copied into proper locations (and packaged if need be) using script ldp_cp
  5. Result seen in the last modified page:

Traditional data content

For docbook - The docbook parser - jade w/DSSSL-stylesheets - is used. It works great. The toolchain is a bit outdated (e.g. not using xslt/xsltproc), but it works wonderfully. If the parser fails badly, the doc goes back to the author. Rarely have that happen.

For linuxdoc - The sgmltools toolchain is used. Again, works just fine. And the same construct goes: if the parser fails badly, the doc goes right back to the author. Rarely have that happen.

Text - no way -- ignored and author informed they need to submit in a given markup. Tell them to subscribe to the discussion list to ask for help, if needed. Not had that occur in years.

Most everything that is submitted is solid, or it's returned. There might be minor tag violations, but it's not enough to worry about, and it's certainly not something that would halt the process.

So? was this hand made?

(Greg) The scripts and putting the tools together? Yes, I did that. Oh, also, part of my steps are to check the file(s) into CVS if the author has not done that. Sometimes, if they use CVS, they tell me where the doc is in the tree (let me know in their submit msg to, I do a cvs update, and run the process. Otherwise, they send the files (again, attached as gzip tar to their message to and I perform the steps I outlined (including checking the files into CVS).

New Wiki Workflow

Write in the wiki, use the "action, Render as docbook" and submit using the traditional chain.

It's necessary to use the exact same name for the HOWTO than the original one (in case of workflow change) to overwrite the old HOWTO.

The render to docbook parser have to be modifyed - may be at each wiki update. The DTD verion must be 4.2, not 4.4 and the images have to have a local image/xxxx.jpg link.  <imagedata fileref="images/avr_html_m7e400112-2.jpg"/> , not  <imagedata fileref=""/>

LDP_data_flow (last edited 2009-03-31 16:49:43 by jdd)