It appears that some directories that should be merged, because the case-insensitive renderings of their names are the same, are missed by the remove-double-dirs.pl script called by 009-case-insensitivity-dirs.sh. One example: when the script concludes, directories "Z3roadstar_1999" and "z3roadstar_1999" remain in $GEO_WORK/geocities/www.geocities.com.
This happens where upper- and lower-case versions of a directory name are non-sequential in $GEO_LOGS/doubles-dir-sorted.txt, which occurs when one directory begins with uppercase, and another with lower. This "double" (or triple, etc.) then isn't caught by remove-double-dirs.pl.
Proposed solution: In $GEO_SCRIPTS/sql/create/find-doubles.sql, change last line to "ORDER BY lowname DESC;" (instead of BY name).
Result: doubles-dir-sorted.txt appears like this, with first-letter-lower and first-letter-upper variants interleaved. E.g.:
/www.geocities.com/~yeata
/www.geocities.com/~Yeata
/www.geocities.com/~Suthernangel
/www.geocities.com/~suthernangel
/www.geocities.com/~minico
/www.geocities.com/~Minico
/www.geocities.com/~mimesq
/www.geocities.com/~MIMESQ
/www.geocities.com/~mcbroom/wish
/www.geocities.com/~McBroom/wish
/www.geocities.com/~McBroom/snows
/www.geocities.com/~mcbroom/snows
/www.geocities.com/~McBroom/sand
/www.geocities.com/~mcbroom/sand
/www.geocities.com/~mcbroom/rositas
/www.geocities.com/~McBroom/rositas
/www.geocities.com/~McBroom/pond
/www.geocities.com/~mcbroom/pond
/www.geocities.com/~McBroom/pictools
/www.geocities.com/~mcbroom/pictools
/www.geocities.com/~mcbroom/images
/www.geocities.com/~McBroom/images
/www.geocities.com/~McBroom/fontinini
/www.geocities.com/~mcbroom/fontinini
/www.geocities.com/~mcbroom/dvs
/www.geocities.com/~McBroom/dvs
/www.geocities.com/~McBroom/dvp
/www.geocities.com/~mcbroom/dvp
/www.geocities.com/~mcbroom/clubpics
/www.geocities.com/~McBroom/clubpics
/www.geocities.com/~McBroom/alps
/www.geocities.com/~mcbroom/alps
/www.geocities.com/~McBroom
/www.geocities.com/~mcbroom
...
It appears that some directories that should be merged, because the case-insensitive renderings of their names are the same, are missed by the remove-double-dirs.pl script called by 009-case-insensitivity-dirs.sh. One example: when the script concludes, directories "Z3roadstar_1999" and "z3roadstar_1999" remain in $GEO_WORK/geocities/www.geocities.com.
This happens where upper- and lower-case versions of a directory name are non-sequential in $GEO_LOGS/doubles-dir-sorted.txt, which occurs when one directory begins with uppercase, and another with lower. This "double" (or triple, etc.) then isn't caught by remove-double-dirs.pl.
Proposed solution: In $GEO_SCRIPTS/sql/create/find-doubles.sql, change last line to "ORDER BY lowname DESC;" (instead of BY name).
Result: doubles-dir-sorted.txt appears like this, with first-letter-lower and first-letter-upper variants interleaved. E.g.:
/www.geocities.com/~yeata
/www.geocities.com/~Yeata
/www.geocities.com/~Suthernangel
/www.geocities.com/~suthernangel
/www.geocities.com/~minico
/www.geocities.com/~Minico
/www.geocities.com/~mimesq
/www.geocities.com/~MIMESQ
/www.geocities.com/~mcbroom/wish
/www.geocities.com/~McBroom/wish
/www.geocities.com/~McBroom/snows
/www.geocities.com/~mcbroom/snows
/www.geocities.com/~McBroom/sand
/www.geocities.com/~mcbroom/sand
/www.geocities.com/~mcbroom/rositas
/www.geocities.com/~McBroom/rositas
/www.geocities.com/~McBroom/pond
/www.geocities.com/~mcbroom/pond
/www.geocities.com/~McBroom/pictools
/www.geocities.com/~mcbroom/pictools
/www.geocities.com/~mcbroom/images
/www.geocities.com/~McBroom/images
/www.geocities.com/~McBroom/fontinini
/www.geocities.com/~mcbroom/fontinini
/www.geocities.com/~mcbroom/dvs
/www.geocities.com/~McBroom/dvs
/www.geocities.com/~McBroom/dvp
/www.geocities.com/~mcbroom/dvp
/www.geocities.com/~mcbroom/clubpics
/www.geocities.com/~McBroom/clubpics
/www.geocities.com/~McBroom/alps
/www.geocities.com/~mcbroom/alps
/www.geocities.com/~McBroom
/www.geocities.com/~mcbroom
...