-*- mode: org -*-

* TODO Algorithm
*** Annahme: n=3 files
*** goal: 1+n pass maximum
***** collect project/file mapping
***** know which project in which files
***** pass 1:
******* file 1: collect all projects with 1 file
******* file 2: collect all projects with 1 file
******* file 3: collect all projects with 1 file
***** pass 2:
******* file 1: collect all projects with 2 files and file=1
******* file 2: collect all projects with 2 files
******* file 3: collect all projects with 2 files
***** pass 3:  
******* file 1: collect all projects with 3 files
******* file 2: collect all projects with 3 files
******* file 3: collect all projects with 3 files
* TODO Algorithm 2
*** Assumptions:
***** n=3 files: TJ, PS, QA
***** simplifications:
******* data of *one* project can be handled in RAM
********* else: stream not on Projects but on Ressources -> probably more passes
******* only very few projects spanned over several organisations (TJ,PS,QA)
********* else: many bucket files, but still ok
*********** it's then ok to kill pass-1 and always write to buckets
************* then we have two other passes:
               1. collect projects in buckets (former pass-2)
               2. add_buckets_to_final()
*********** if not: other approach with 1+n passes
************* 1 pass: collect projects per file
************* then 1 pass foreach organisation, each pass over all 3 files
************* result 4 passes vs. 2 passes
*** goal: 2 passes maximum
***** 1st pass: collect which projects in which files
***** 2nd pass: go through all 3 files, write final+buckets, write buckets to final
*** pass 1:
***** foreach files
******* xml::twig::callback_Project
********* $PROJECT_FILE-Hash: "projectID+file"++
*** pass 2:
***** XML startblock
***** foreach files
******* callback_Project
********* if keys $PROJECT_FILE-Hash{$project}{files} > 1
*********** add_project_to_bucket($project)
********* else
*********** add_project_to_final($project)
***** add_buckets_to_final()
***** XML endblock
*** functions:
***** add_project_to_final:
******* $project->flush(FINAL)
***** add_project_to_bucket:
******* open PROJECTFILE, ">>", "$projectID.bucket"
******* $project->flush(PROJECTFILE)
******* close PROJECTFILE
***** add_buckets_to_final:
******* foreach *.bucket
********* open(PROJECTFILE)
********* $project = new XML_Element("Project")
********* xml::twig::callback_Project:
*********** $project.set_att("name",      $cur.att("name")) unless $project.attr("name");
*********** $project.set_att("projectID", $cur.att("name")) unless $project.attr("projectID");
*********** foreach $cur.ressources:
************* $project.add($resource)
********* add_project_to_final($project);
*** class:
* TODO Bugs
*** TODO Resource elements twice inside Project
* TODO Project/CustomInformation?
*** what if a project is spanned over several organisations?
***** <Resource> elements are appended into one <Resources>
***** what to do with all other elements besides the several <Resource> elements
*** alternatives:
***** ignore completely, throw away
***** take all, by adding them all to surrounding (single) <Project>
***** take whole <Project> element from first (unsorted, any) <Project>
*** multiple?
*** take first?