Last edited on 2009-02-01 22:47:22 by stolfi

The "fragments" project
92 - general cleanup of IAB-2002 and IAB-2002-unicamp on 2009-01-31

  We had two old versions of the RAAB site, with
  mod-dates clustered around two disinct epochs.
  In directory "projects/fragments/IAB-2002":
  
    "2003-03-10-manaus" (~ 2003-03-10)
    "2009-01-30-manaus" (~ 2003-07-20)
    
  To simplify the comparison, we did some cleanup in
  "2003-03-10-manaus":
  
  See 2003-03-10-manaus/91-Notebook.txt
    
  As part of that rearrangement we moved some files
  devoted to alignment to a new directory "2003-02-07-align".
  
  We also moved files related to segmentation from
  "2003-03-10-manaus" to "2003-02-07-segm"
  
  Comparing the directory structures:
    
    for s in {2003-03-10,2009-01-30}-manaus ; do 
      ( cd $s && \
        find {PUB,TEC,ETC,PRE}/ -type d -name '[0-9][0-9][0-9]' -print \
          | sort | uniq \
          > .all-NNN
      ) ;
    done
    prdiff -Bb {2003-03-10,2009-01-30}-manaus/.all-NNN > .diff-NNN
    
  Only some */misc/* were missing in 2003-03-10-manaus.  Those
  had no fragments or other interesting files, only links to
  the basic images.
  
  Checking consistency of image names:
  
    for s in {2003-03-10,2009-01-30}-manaus ; do 
      ( cd $s && \
        find {PUB,TEC,ETC,PRE}/ -name '*.jpg' -print \
          | sed -e 's:-[a-z][.]jpg$::' -e 's:[.]jpg$::' \
          | sort | uniq \
          > .all-images 
      ) ;
    done
    prdiff -Bb {2003-03-10,2009-01-30}-manaus/.all-images > .diff-images
    
  Again, only "misc" images are missing from 2003-03-10-manaus.
  
  Comparing the various coordinate files:

    for e in fctrs grays grpts balls ; do
      for s in {2003-03-10,2009-01-30}-manaus ; do 
        ( cd $s && \
          find {PUB,TEC,ETC,PRE}/ -name "*.${e}" -print \
            | sed -e 's:[.]'"${e}"'$::' \
            | sort | uniq \
            > .all-${e}
        ) ;
      done
      prdiff -Bb {2003-03-10,2009-01-30}-manaus/.all-${e} > .diff-${e}
      wc -l .diff-${e}
    done

  There were several differences in the distrbution of ".fctrs" files.

      Only in 2009-01-30-manaus:
      ./PRE/mirr/004
      ./PRE/mirr/006
      ./PRE/mirr/008
      ./PRE/mirr/009
      ./PRE/mirr/011
      ./PRE/mirr/015
      ./TEC/mirr/021
      ./TEC/mirr/023
      ./TEC/mirr/025
      ./TEC/mirr/026
      ./TEC/mirr/027
      ./TEC/mirr/028
      ./TEC/mirr/029
      ./TEC/mirr/030
      ./TEC/mirr/031
      
      ./PUB/misc/000
      ./TEC/misc/005
      ./TEC/misc/010
      ./TEC/misc/011
      ./PUB/misc/020
      ./PUB/misc/021
      
      ./ETC/mult/001
      ./TEC/mult/001
      ./TEC/mult/002
      ./TEC/mult/003
      ./TEC/mult/004
      ./TEC/mult/005
      
      ./PUB/objs/001
      ./ETC/objs/011
      ./PUB/objs/011
      ./PUB/objs/014
      ./PUB/objs/019
      
      ./PRE/stds/000
      ./PRE/stds/001
      ./TEC/stds/018
      ./TEC/stds/021
      ./TEC/stds/028
    
  Many files occurred in both versions, but with different dates and
  slightly different contents. For instance, the "2009-01-30-manaus" versions
  of the "*-n.jpg" files, showing the fragment numbers, have smaller
  yellow dots, and bottom-aligned instead of middle-aligned labels.
  The "*-g.jpg" images, that show the ideal grids fitted to the
  printed grid images, also seem to be better positioned in the
  "2009-01-30-manaus" version.
  
  There were differences in the "*.[gpfyb]cmds" files but these
  are derived files.  The "*.grraw" files were present only in "2009-01-30-manaus"
  but were found to be identical to the "*.grpts" files and deleted.

  Comparing the renaming tables:
  
    prdiff -Bb {2003-03-10,2009-01-30}-manaus/raw-to-sorted.tbl
    
  There were only two cosmetic differences in comments, so:
  
    rm -v 2003-03-10-manaus/raw-to-sorted.tbl
    
  At this moment (1009-01-31) the 2003-03-10-manaus is junk.