Paul Roeland: The value of longevity

published Nov 29, 2024

Talk by Paul Roeland at the Plone conference 2024 in Brasilia.

Link to talk information on Plone conference website.

I work for the Clean Clothes Campaign: https://cleanclothes.org/

After three large disasters in factories in 2012 and 2013 with over 1000 deaths, it took three years to get an agreement with clothes manufacturers to get 30 million dollar compensation. It does not bring lives back, but it helps the survivors.

See Open Supply Hub for open data that we collected, for checking which brands are produced in which factories.

Documenting history matters. Stories must be told.

The global closing industry is worth around 1.8 trillion dollars, in a country that would put them on the 12th place in the world. 75 million workers.

Our strongest weapon: backlinks. We have links from OECD, UN, wikipedia, school curriculum, books. Especially those last two don't change ever, so you should never change urls.

Plone: enable the sitemap, please, why not by default? Create a good robots.txt. I weekly check Google Search console, looking for broken links. Tag early, tag often, great tool, even if you have an AI do it.

Our website: started 1998 written in Notepad, 2004 Dreamweaver, 2006 Bluefish, 2010 Joomla, 2013 Plone 4, 2020 Castle CMS (opinionated distribution of Plone, but does not really exist anymore) 2024 Plone 6 with Volto Light Theme (work in progress). Thank you kitconcept for all the help, especially Jonas.

Migrations are painful. Along the years we used wget to csv to SQL to csv, Python script, "Franken-mogrifier", collective.exportimport.

Lessons learned: stable urls are awesome, migrations are painful. Please don't try to salvage CSS from your old site, just start fresh in your new system. Do not try to migrate composite pages or listings.

What if your website does not provide an export? Use wget, still works and is better than httrack. sed/awk/regex are your friend. archivebox (WARC).

Document your steps for your own sanity.

To manage json, jq or jello can be used. sq is a Swiss knife for json/sql/csv. emuto is a hybrid between jq and GraphQL.

Normalize import/export. We have `plone.exportimport` in core now.

In the future I would like a plone exporter script that accepts a regex and exports only matching pages. Switch backends: ZODB, relstorage, nick, quantum-db. Sitewide search/replace/sed. Sneakernet is useful in difficult countries where you cannot send data over the internet: so export to a usb stick.

A backup is only a backup if it regularly gets restored so you know that it works.

  • Keeping content and URL stability is a superpower.
  • Assuming that export/import/backup/restore/migration are rare occurrences, is wrong.
  • Quick export/import is very useful.

Do small migrations, treat it as maintenance. Don't be too far behind. Large migrations one every five years will be costly. Do a small migration every year. Do your part. Clients should also do their part, by budgeting this yearly. That is how budgeting works. Use every iteration to review custom code.

Make your sites live long and prosper.