TODO

• URL change of Sphinx docs to be more google friendly

  • http://dayabay.phys.ntu.edu.tw/h/ to http://dayabay.phys.ntu.edu.tw/heprez/

INTEGRATION

AFS access

• try regain AFS via IHEP : sent to maqm
• try regain AFS via SLAC : sent to Tim

Mars has AFS

[mars] /home/hep/blyth > export CVSROOT=/afs/slac.stanford.edu/www/xorg/hfag/cvs
[mars] /home/hep/blyth > cvs checkout Papers
cvs [checkout aborted]: /afs/slac.stanford.edu/www/xorg/hfag/cvs/CVSROOT: Permission denied
[mars] /home/hep/blyth >

URGENT DEVELOPMENT

decouple fast from slow scraping

Decouple fast text html,tex,svg pulling out of queries from slow conversions to pdf,jpg,png

1. skip jpg rendition of the PDF compilation
2. skip all binary conversions on query pass, do these (or omit where possible) in separate pass
3. combination page latex is not useful without flattening : so skip and just retain PDF link ?
4. exist console full of convert errors failing to create jpg from ps heprez:ticket:177

NON URGENT DEVELOPMENT

enhancements

1. more prominent link to http://www.slac.stanford.edu/xorg/hfag/
2. find home for reference links such as http://www.xquery.co.uk/tutorial_xquery_get_started/get_tutorial_page/xquery_global_variables
3. fix rezdb previews for multi factor qwn (maybe with MathJax)
4. investigate clientside javascript possibilities : MathJax and JSXChart
5. cache kicks in only on 3rd pass ticket:178
6. suppress non-contributors in tables (and plots?)
7. automated charge conservation checking ? can the charge be systematically be determined from PDG code without too much pain ? or is that in pdg.xml ?
8. factor out latex2html conversions in the automated workflow, this was manually done post v007 end_of_2011
   1. primary issue is if the metrics from the latex2html conversion are still needed ?

digraph foo { "tex" -> "pdf"; "tex" -> "png" [label=latex2html, color=red] ; "pdf" -> "svg"; "svg" -> "opdf"; "svg" -> "opng" [label=illustrator, color=green] ; }

1. longer term want to move away from commercial illustrator (eg to inkscape)
2. Trac RST previews fail to handle the Sphinx toctree directive, eg
   • some Trac config switch to get rid of RST previewing ?
   • http://dayabay.phys.ntu.edu.tw/tracs/heprez/browser/trunk/qxml/index.rst
   • http://trac.edgewall.org/ticket/9915 takes approach of making error messages less objectionable,
   • I would prefer presentation of source RST as plain text, without renaming all .rst to .txt

recursion inversion

1. current hidden approach : inflexible, difficult to optimize, difficult error handling
2. do recursion at top level : quickly yield combination html (hmm maybe move to xhtml to avoid having to have xml fragments + final html) then recurse over links within that and so on
3. this differs from current arrangement where it the comb page waits on it dependents
4. compilation latex relies on having run scape and created already all the .tex : somewhat wrong ordering for setting shrinkages

qxml : dbxml based alternative to eXist

1. qxml : make which maps to load cli controllable
2. exist2qxml.py : shakedown selective running, maybe getting interference with map loading ?
3. exist2qxml.py : metadata checking, seeing some suprising timestamps reported, in the ballpark but hours off
4. look into qxml scrape based off transfered /db/hfagc_prod into avg.dbxml : ie retain exist avg but try to benefit from qxml speed for the scrape
5. mode presentation uses the product particle ordering from the first corresponding mode with the qtag
   1. this is not reproducible in dbxml due to node ordering differneces (eg 9 qtags in qtag2xml have different presentation)
   2. this technique is liberally sprinkled, not so easy to change even if an approach can be identified
6. xquery module renaming and reorganisation, rezu rezt rezx meanings have been lost due to difficult exist development environment

misc

• DONE : exist-up-hfagc warns WILL NEED exist-place-xsl but exist-place-xsl sees nothing to do, change it to be based on diffs r:776

• DONE : hfag, suspect that a reboot will lead to the wrong apache getting started .. cleaned up init.d

• DONE : table 72, overlap http://hfag.phys.ntu.edu.tw/b2charm/00109.html

• DONE : missing key table : suspect r759 the 99999 collection not created

• DONE : broken table link in key caption due to incorrect hardcoded last table following classification change:

  * ``table.xq`` is source for the key caption, changed::
       \ref{tab:hfc03300} to \ref{tab:hfc04300}

• DONE : Yasmine in the : “who we are”

  1. /db/hfagc_system/members.xml is sourced from ~/heprez/hfag/mods/webapp/hfagc/system/members.xml
  2. propagated by heprez-prepare–/exist-prepare

within 00104, one LHCb result appears twice (FIXED)

• cms01:b2charm/end_of_2011/BR_-521_-211+443xoBR_-521_-321+443.html
• cms01:b2charm/end_of_2011/1204030.html
• cms01:b2charm/end_of_2011/1204010.html

pig-in-poke issue, implicated with qwn mis-classification

cms01:b2charm/end_of_2011/00108.html heprez:ticket:171

• also range based plot splitting involved
• DONE : belle result mis-classification implicated, so get that fixed before investigating further
• DONE : fixed up error SVG to avoid remedial simon:20120417 blyth$ mv ./hfagc/00108/01128/01128.svg ./hfagc/00108/01128/01128.svg.dud
• cms01:b2charm/end_of_2011/01128.svg

SUPERCEEDED SINGLE REZ VISIBLE (FIXED v006)

Superceeded results should be visible, only “inactive” are hidden,

• http://cms01.phys.ntu.edu.tw/b2charm/0905018.html
• rez.xq invokes rezt:present-rez, include superceeded in r:775

echo "distinct-values(collection('rem')//rez:status)" | qxml -

active
inactive
superceeded