Libraries to group together formulas, parameter sets, maps, L-system definitions, IFS definitions, etc.

[LegalizeAdulthood]

Ever since I created an installer for Iterated Dynamics that put the various data files in Program Files (a read-only location), there's been a struggle to identify locations of files for reading vs. locations for writing. I haven't really been happy with the situation so far, where each type of file has it's own bespoke logic for identifying the read location.

In issue #95 I proposed specifying a list of directories for the various file types to assist in searching for files to read. Unfortunately this just makes a complex situation already more complicated as each type of file would be searching through possibly different directory locations.

It's also the case that these various files are related to each other. Parameter sets can refer to formula files, IFS files, L-system files, colormap files. Looking forward down the roadmap, there will be cases where formula files will reference other files following the syntax of Ultra Fractal.

So, instead of having individual search locations for each type of file, I'm proposing the idea of a Library. A Library is simply a directory containing subdirectories for each file type organized similar to the home directory in the source tree. Instead of specifying a list of search locations per file type, you specify a list of library locations to search. The search algorithm will be roughly as follows:

1. For each library location
   1. Look in the appropriate subdirectory for the file type (e.g. maps, pars, etc.) for the file.
   2. Look in the library directory itself for the file
2. Report an error if not found in any known library locations.

For saving files (writing PAR sets, batch files, color maps, etc.) there will be a single library path and again, the appropriate subdir will be selected. For saving images, the subdirectory will be images under the library save location. When saving, the appropriate subdirectory for the type of file being written will be created if it doesn't exist.

Currently the default save location (on Windows) is the user's Documents folder. This implies that subdirectories would be created in the user's documents folder when saving files. Instead of putting everything in the documents folder, the default save location should be a subdirectory named Id of the documents folder so that everything is grouped together into a single folder. Since the user's documents folder is not a fixed path, the convention shall be that setting the value of the save library location to a relative path will always be relative to the user's documents folder and will default to Id.

For reading, wxWidgets can treat a ZIP file as a file system, so it should be possible to specify read locations as ZIP files, making it easy to share libraries between users.

For the purposes of this discussion, I'd like to hear your opinions on this idea and see if I've missed anything. I think this will be a better solution than having a plethora of search locations for the different file types.

[PaulTheLionHeart]

In ManpWIN, I have a ManpWIN.ini file that lives in AppData\Roaming\ManpWIN folder. This saves all the preferences that are set up in the Setup menu as well as the default locations for all the file types. as well as the most recently used filename of each type. This file is recreated every time ManpWIN exits.

;========================================================================
; Configuration File for Paul's Fractals: © Paul de Leeuw 25/02/2020
;========================================================================
;
; Note the config lines must start at character 0 and are case sensitive
;
; Comments are started with a ; character
;
;
; Path for PNG files
PNGPath=C:\Users\Paul\AppData\Roaming\ManpWIN\png
; Path for GIF files
GIFPath=C:\Users\Paul\AppData\Roaming\ManpWIN\gif
; Path for MPG files
MPGPath=C:\Users\Paul\AppData\Roaming\ManpWIN\mpg
; Path for PNG files from animation generator
ANIMPNGPath=C:\Users\Paul\AppData\Roaming\ManpWIN\animpng
; Path for COL files
COLPath=C:\Users\Paul\AppData\Roaming\ManpWIN\col
; Path for MAP files
MAPPath=C:\Users\Paul\AppData\Roaming\ManpWIN\map
; Path for SCI files
SCIPath=C:\Users\Paul\AppData\Roaming\ManpWIN\sci
; Path for PAR files
PARPath=C:\Users\Paul\AppData\Roaming\ManpWIN\par
; Path for KFR files
KFRPath=C:\Users\Paul\AppData\Roaming\ManpWIN\kfr
; Path for Fractint PAR files
FracPARPath=C:\Users\Paul\AppData\Roaming\ManpWIN\frantpar
; Path for IFS files
IFSPath=C:\Users\Paul\AppData\Roaming\ManpWIN\ifs
; Path for Lsys files
LSYSPath=C:\Users\Paul\AppData\Roaming\ManpWIN\lsys
; Path for Formula files
FRMPath=C:\Users\Paul\AppData\Roaming\ManpWIN\frm
;
; PNG file
PNGFile=Manp-230324-160150.png
; COL file
COLFile=Sunshine.COL
; MAP file
MAPFile=Colour info from Fractint Par File
; SCI file
SCIFile=TestSmoothing.sci
; Fractint PAR file
FracPARFile=C:\Users\Paul\AppData\Roaming\manpWIN\frantpar\2007.08.12-Straight_Forward.par
; PAR file
PARFile=SineDerSlope01.par
; KFR file
KFRFile=Desert.kfr
; IFS file
IFSFile=C:\Users\Paul\AppData\Roaming\manpWIN\ifs\MANPWIN.IFS
; Lsys file
LSYSFile=C:\Users\Paul\AppData\Roaming\manpWIN\lsys\FRACTINT.L
; Formula file
FRMFile=C:\Users\Paul\AppData\Roaming\manpWIN\frm\fract196.FRM
;
; Do we want to zoom along the edge?
;
ZoomEdge=FALSE
;
; Do we want to use Fractint Palette?
;
UseFractintPalette=TRUE
;
; Allow MPEG generator to dump stats into a stats.txt file
;
AllowMPEGStats=TRUE
;
; Use the default directory structure in user data area
;
UserPath=TRUE
;
; Number of threads used in multi-threaded perturbation only.
; If Number of threads = 0, then non-threaded perturbation is used
;
ThreadNumber=0
;
; Do we display a palette on the right side of the image?
;
DisplayPalette=TRUE
;

[LegalizeAdulthood]

Id currently has the sstools.ini and id.cfg files for storing some state that is reloaded every time Id starts. This is somewhat antiquated (as are .ini files in general), but it is the existing mechanism.

The issue here isn't how initial values are loaded. The issue is how data files are organized, searched and their save locations are determined.

There are tons of ZIP files out there containing various parameter sets, IFS definitions, etc. As a user I don't want to mingle random stuff from the internet with the stock definitions that ship with Id or with my own definitions that I've created. Libraries let me keep them separate but still locate them easily.

There are some UI changes that have to go with this as well because the existing browse screens simply assume a single directory and there's no explicit way to manage libraries.

[LegalizeAdulthood]

Phew! Well that took a whole bunch of code massaging, but now all reading and writing of files is done via the libraries mechanism and the documentation has been updated to describe libraries and how they work.

Unit tests were added for the search mechanisms and all the file-based tests were updated with the new file locations on output.

[PaulTheLionHeart]

Well done Richard.