I recently bought an UPS by the Innova company.
On the box it was written “Linux”. I opened the box and got a CD disk. On the CD was the folder “linux”. In the folder was some old broken version of the program Upsilon and no driver. (The built-in driver coming with Debian Linux also didn’t work.)
I returned the UPS to the seller.
Shame to the Innova company.
I’ve copied my
/var/www, MySQL database, and
/etc from a sever to an other server.
After this I found that my MySQL does not work anymore.
The reason for the trouble was that
/etc/mysql/debian.cnf contains (twice) an auto-generated password for the database, and so after copying my old
/etc directory from my old server the password didn’t match the new password stored in my MySQL database.
The solution: Do not overwrite
/etc/mysql/debian.cnf file when copying
/etc directory, but leave it as it was autogenerated by the new server installation script.
Note: This is a about Debian Linux.
My proposal for a new data description language NLang alternative to RDF is probably deadborn.
I have missed the rationale:
The rationale for NLang was that the data is not messed up when loading information from several files.
But today I’ve realized that I can just write RDF software in such a way that information from several files that they won’t intermix.
So I can continue my work with traditional RDF. There is no real need to create that fancy NLang.
There is RDF. Why to create a new data description language? Several RDF files (possibly downloaded from the Web and non-trusted) may be merged in such a way that it creates contradictory data. It contradicts to design goals of this my project.
We need to create a language which describes objects and objects are in some sense immutable (cannot be changed by loading one more file with other objects).
With every IRI they are associated one or more objects. Every object has some (say N) properties.
So I propose a language to describe objects and propose to call it NLang (because unlike RDF objects may have N>1 properties).
Next it follows an example (similar but different to Turtle syntax) after which I am going to model the syntax:
a :transformer ;
dc:description <http://...> ;
# Other Dublin Core metadata.
:source-namespace <...> ;
:target-namespace <...> ;
:precedence <...> ;
:scripts ref <http://example.org/scripts>
a :xslt ;
:version "2.0" ;
:script-url <http://example.org/scripts/foo.xslt> ;
:transformer-kind :entire ;
:name "debug" ;
:name "other" ;
#:initial-context-node ... ; # See XSLT 2.0 spec.
initial-template "first" ;
initial-mode: "first" ;
completeness 0.9 ;
stability 0.9 ;
As an update to the previous version of the specification I have written a new draft of Automatic transformation between XML documents with different namespaces.
This updated version supports the choice of the user to download only information referenced from the source or only information referenced from the destination.
For example, it allows, when directly transforming from an XML namespace A to an XML namespace B, to download only transformation information at A or only information at B (provided that this data is available at A or at B respectively).