This reverts commit a6b189b. Conflicts: tagfs.lc
This reverts commit 1286aad.
- g_list_union sorts on its own
- uses `phases' of substitution that were built into the execution sequence in lex.pl
- Log system now makes use of macros for file-based log level. This allows the same level of granularity and flexiblity as before, but without the troublesome use of behind-the-scenes assignment of global variables * Also added a command line option to specify the log file name * Removed references to _log_level outside of log.c (it was renamed anyway) - The root directory is now detected - Commented out references to search_fs in makefile and in tagfs.lc (uses some older stuff) - Turned references to a lex.pl-generated tagfs_operations struct into a macro %%%operatons_struct_name%%% for better separation - Rewrote (proc_options => process_options) to use the glib option parser. Uses global config variables (prefixed with c_). - Removed storage of the mount directory in our tagfs_data; it makes processing options less stupid and we didn't need to store it anyway.
- Removed an over-prominent log message in tagdb_util.c - Deleted legacy code in tagfs.lc
- Moved tagdb.fs to tagdb_fs.lc - Removed superfluous logging functions in file_log.c - Added file_log, tagdb_fs, fs_util to the build - Fixed `file_info_read' header - Cleaned up lex.pl * In SUBFS: * function abstraction * corrected creation of subfs_component struct * removed the "auto-registration" block--this can come back later with a better implementation * removed some fold markers * made the file basename more explicitly global -- `g_' prefix and more specific name - commented out log_level check in log.c * Needs to be cleaned up...probably should use fewer macros... - changed subfs_component back to containing a fuse_operations struct. holding a pointer doesn't work as nicely for static initialization - added subfs to main and subfs actually registers the components - deleted `proc_options' in the main file. Will replace it with glib's option parser soon
- Changed `chug_path' into an actual function * `g_alloca' is used for allocation in all of the places its called, so be careful of a stack overflow (fuse gives us more restriction there) - renamed tagfs.l to tagfs.lc - added a tagdb_new function for when you don't want to load in a database
- Lots of things changed and files were added - Does Not Build
- Extracted out some procedures that might be used by multiple subfs components
- We sacrificed the parameter names, so logs are somewhat less clear - This is still incomplete since it depends on changes in the subfs system
- Build process (INCOMPLETE) detailed in makefile.subfs and lex.pl - *.fs files are processed by lex.pl to get .fc files which compile to .fo, mirroring the path for query classes
- A lot more is being written on my blog about this, so I might not finish the notes in here
- The listen file should have the contents of the last query made by _writing_ to the file which contents must persist until a new query is made
- Query files now have a .q extension and turn into .qc through lex.pl and finally query_<base>.o when compiled - Removed the underscore from the $base variable - Added auto-generation of the subfs_component struct