20110806 ======== Upload files with non-ascii names --------------------------------- This week I tried to understand why in Eupen we cannot upload files with non-ascii characters in their name. I added a new file :srcref:`/lino/tests/misc/diag.py` (converted from the older ``timtools diag`` command). On Jana this didn't even run because there were no locales installed: the value of `LANG` was simply `"C"`; typing the :kbd:`ö` key on my Estonian keyboard produced some strange effects, but not a letter "ö". The solution was simple (now that I found it):: sudo dpkg-reconfigure locales This was probably because Jana was installed from some Proxmox's default Debian template. New application "Concise Reference Labels" ------------------------------------------ I started a new application :mod:`lino.apps.crl`, a kind of `personal information manager `_ for a customer who has several good reasons to get a free pilot project: he is both patient and perfectionist; he has some rather exclusive wishes including a few challenges; he has an existing application that needs to be converted... and he's my father. My father has worked in several libraries, classification is one of his hobbies, and although he recently got 78 years old he doesn't stop advocating a labeling system he discovered, used and developed throughout his life. In September 2009 he published an article `De l'ordre dans les idées `_ about it. He didn't yet worry to find a name for this labeling system. So I'll call it **Concise Reference Labels**. It is based on the observation that printed books in a library have limited space for labels. Now imagine a classical library, where humans directly manipulate lots of books and need to find the right place for each. In this situation it is useful to have a codification system that is easy to learn and to remember at least for the people who use it every day. The classical approaches in the western world for this are `UDC `_ and `DDC `_. They both have one disadvantage: they don't use the full range of "rememberable characters" and thus need relatively much space on a label. My father's CRL approach assumes that people accept the need of a *localized classification subset* and are willing to do the work of discussing and following such a subset.