X-Git-Url: http://pere.pagekite.me/gitweb/homepage.git/blobdiff_plain/89da5b96e7f055a55c0dd5629a152fcad71f643f..271aba3cc2efb48d8c0720a80a64f7845124cba6:/blog/archive/2019/03/index.html diff --git a/blog/archive/2019/03/index.html b/blog/archive/2019/03/index.html index 09f98f5ea5..c181d1b2e3 100644 --- a/blog/archive/2019/03/index.html +++ b/blog/archive/2019/03/index.html @@ -21,6 +21,231 @@

Entries from March 2019.

+
+
+ PlantUML for text based UML diagram modelling - nice free software +
+
+ 25th March 2019 +
+
+

As part of my involvement with the +Nikita +Noark 5 core project, I have been proposing improvements to the +API specification created by The +National Archives of Norway and helped migrating the text from a +version control system unfriendly binary format (docx) to Markdown in +git. Combined with the migration to a public git repository (on +github), this has made it possible for anyone to suggest improvement +to the text.

+ +

The specification is filled with UML diagrams. I believe the +original diagrams were modelled using Sparx Systems Enterprise +Architect, and exported as EMF files for import into docx. This +approach make it very hard to track changes using a version control +system. To improve the situation I have been looking for a good text +based UML format with associated command line free software tools on +Linux and Windows, to allow anyone to send in corrections to the UML +diagrams in the specification. The tool must be text based to work +with git, and command line to be able to run it automatically to +generate the diagram images. Finally, it must be free software to +allow anyone, even those that can not accept a non-free software +license, to contribute.

+ +

I did not know much about free software UML modelling tools when I +started. I have used dia and inkscape for simple modelling in the +past, but neither are available on Windows, as far as I could tell. I +came across a nice +list +of text mode uml tools, and tested out a few of the tools listed +there. The PlantUML tool seemed +most promising. After verifying that the packages +is available in +Debian and found its +Java source under a GPL license on github, I set out to test if it +could represent the diagrams we needed, ie the ones currently in +the +Noark 5 Tjenestegrensesnitt specification. I am happy to report +that it could represent them, even thought it have a few warts here +and there.

+ +

After a few days of modelling I completed the task this weekend. A +temporary link to the complete set of diagrams (original and from +PlantUML) is available in +the +github issue discussing the need for a text based UML format, but +please note I lack a sensible tool to convert EMF files to PNGs, so +the "original" rendering is not as good as the original was in the +publised PDF.

+ +

Here is an example UML diagram, showing the core classes for +keeping metadata about archived documents:

+ +
+@startuml
+skinparam classAttributeIconSize 0
+
+!include media/uml-class-arkivskaper.iuml
+!include media/uml-class-arkiv.iuml
+!include media/uml-class-klassifikasjonssystem.iuml
+!include media/uml-class-klasse.iuml
+!include media/uml-class-arkivdel.iuml
+!include media/uml-class-mappe.iuml
+!include media/uml-class-merknad.iuml
+!include media/uml-class-registrering.iuml
+!include media/uml-class-basisregistrering.iuml
+!include media/uml-class-dokumentbeskrivelse.iuml
+!include media/uml-class-dokumentobjekt.iuml
+!include media/uml-class-konvertering.iuml
+!include media/uml-datatype-elektronisksignatur.iuml
+
+Arkivstruktur.Arkivskaper "+arkivskaper 1..*" <-o "+arkiv 0..*" Arkivstruktur.Arkiv
+Arkivstruktur.Arkiv o--> "+underarkiv 0..*" Arkivstruktur.Arkiv
+Arkivstruktur.Arkiv "+arkiv 1" o--> "+arkivdel 0..*" Arkivstruktur.Arkivdel
+Arkivstruktur.Klassifikasjonssystem "+klassifikasjonssystem [0..1]" <--o "+arkivdel 1..*" Arkivstruktur.Arkivdel
+Arkivstruktur.Klassifikasjonssystem "+klassifikasjonssystem [0..1]" o--> "+klasse 0..*" Arkivstruktur.Klasse
+Arkivstruktur.Arkivdel "+arkivdel 0..1" o--> "+mappe 0..*" Arkivstruktur.Mappe
+Arkivstruktur.Arkivdel "+arkivdel 0..1" o--> "+registrering 0..*" Arkivstruktur.Registrering
+Arkivstruktur.Klasse "+klasse 0..1" o--> "+mappe 0..*" Arkivstruktur.Mappe
+Arkivstruktur.Klasse "+klasse 0..1" o--> "+registrering 0..*" Arkivstruktur.Registrering
+Arkivstruktur.Mappe --> "+undermappe 0..*" Arkivstruktur.Mappe
+Arkivstruktur.Mappe "+mappe 0..1" o--> "+registrering 0..*" Arkivstruktur.Registrering
+Arkivstruktur.Merknad "+merknad 0..*" <--* Arkivstruktur.Mappe
+Arkivstruktur.Merknad "+merknad 0..*" <--* Arkivstruktur.Dokumentbeskrivelse
+Arkivstruktur.Basisregistrering -|> Arkivstruktur.Registrering
+Arkivstruktur.Merknad "+merknad 0..*" <--* Arkivstruktur.Basisregistrering
+Arkivstruktur.Registrering "+registrering 1..*" o--> "+dokumentbeskrivelse 0..*" Arkivstruktur.Dokumentbeskrivelse
+Arkivstruktur.Dokumentbeskrivelse "+dokumentbeskrivelse 1" o-> "+dokumentobjekt 0..*" Arkivstruktur.Dokumentobjekt
+Arkivstruktur.Dokumentobjekt *-> "+konvertering 0..*" Arkivstruktur.Konvertering
+Arkivstruktur.ElektroniskSignatur -[hidden]-> Arkivstruktur.Dokumentobjekt
+@enduml
+
+ +

The format is quite +compact, with little redundant information. The text expresses +entities and relations, and there is little layout related fluff. One +can reuse content by using include files, allowing for consistent +naming across several diagrams. The include files can be standalone +PlantUML too. Here is the content of +media/uml-class-arkivskaper.iuml:

+ +
+@startuml
+class Arkivstruktur.Arkivskaper  {
+  +arkivskaperID : string
+  +arkivskaperNavn : string
+  +beskrivelse : string [0..1]
+}
+@enduml
+
+ +

This is what the complete diagram for the PlantUML notation above +look like:

+ +

+ +

A cool feature of PlantUML is that the generated PNG files include +the entire original source diagram as text. The source (with include +statements expanded) can be extracted using for example +exiftool. Another cool feature is that parts of the entities +can be hidden after inclusion. This allow to use include files with +all attributes listed, even for UML diagrams that should not list any +attributes.

+ +

The diagram also show some of the warts. Some times the layout +engine place text labels on top of each other, and some times it place +the class boxes too close to each other, not leaving room for the +labels on the relationship arrows. The former can be worked around by +placing extra newlines in the labes (ie "\n"). I did not do it here +to be able to demonstrate the issue. I have not found a good way +around the latter, so I normally try to reduce the problem by changing +from vertical to horizontal links to improve the layout.

+ +

All in all, I am quite happy with PlantUML, and very impressed with +how quickly its lead developer responds to questions. So far I got an +answer to my questions in a few hours when I send an email. I +definitely recommend looking at PlantUML if you need to make UML +diagrams. Note, PlantUML can draw a lot more than class relations. +Check out the documention for a complete list. :)

+ +

As usual, if you use Bitcoin and want to show your support of my +activities, please send Bitcoin donations to my address +15oWEoG9dUPovwmUL9KWAnYRtNJEkP1u1b.

+ +
+
+ + + Tags: english, nice free software, noark5, nuug, standard. + + +
+
+
+ +
+ +
+ 24th March 2019 +
+
+

Yesterday, a new release of +Nikita +Noark 5 core project was +announced +on the project mailing list. The free software solution is an +implementation of the Norwegian archive standard Noark 5 used by +government offices in Norway. These were the changes in version 0.3 +since version 0.2.1 (from NEWS.md):

+ +
    +
  • Improved ClassificationSystem and Class behaviour.
  • +
  • Tidied up known inconsistencies between domain model and hateaos links.
  • +
  • Added experimental code for blockchain integration.
  • +
  • Make token expiry time configurable at upstart from properties file.
  • +
  • Continued work on OData search syntax.
  • +
  • Started work on pagination for entities, partly implemented for Saksmappe.
  • +
  • Finalise ClassifiedCode Metadata entity.
  • +
  • Implement mechanism to check if authentication token is still + valid. This allow the GUI to return a more sensible message to the + user if the token is expired.
  • +
  • Reintroduce browse.html page to allow user to browse JSON API using + hateoas links.
  • +
  • Fix bug in handling file/mappe sequence number. Year change was + not properly handled.
  • +
  • Update application yml files to be in sync with current development.
  • +
  • Stop 'converting' everything to PDF using libreoffice. Only + convert the file formats doc, ppt, xls, docx, pptx, xlsx, odt, odp + and ods.
  • +
  • Continued code style fixing, making code more readable.
  • +
  • Minor bug fixes.
  • + +
+ +

If free and open standardized archiving API sound interesting to +you, please contact us on IRC +(#nikita on +irc.freenode.net) or email +(nikita-noark +mailing list).

+ +

As usual, if you use Bitcoin and want to show your support of my +activities, please send Bitcoin donations to my address +15oWEoG9dUPovwmUL9KWAnYRtNJEkP1u1b.

+ +
+
+ + + Tags: english, noark5, nuug, offentlig innsyn, standard. + + +
+
+
+