Thanks Codaz, this is exactly the feedback I was looking for! Let's see :
Only the GPSFS ( and NOT! the GPSFS1/2/2/3/4 ) files have the information on the tiles. GPSFS1/2/3/4 files simply contain the image dumps.
At the moment I am not doing any super extra error checkings to see if you "obeyed" the basic rule of "Select the GPSFS file only"...
I tried this several times and saw no freezing. Altough it takes quite some time for windows to delete/reinit the list box, especially if there were throusands of entries. Be patient...
Level 0 is the root map. It does not make sense deleting it. It also is the smallest, only 4 tiles.
Empty maps do not occupy any image space, but are important for map logistics. They only occupy a mere 8 bytes each in GPSFS.
I'll look into that. Why would you want to select all tiles?
It is very likely a PNG based map. v0.1 does not have PNG-recompression capability. It will recompress PNG based maps in JPG. To have roughly the same image quality, it would require that many bytes. I will fix this in probably v0.3 or so. Meanwhile, I found that if you go around 70~80% of your original PNG based files, it still produces a pretty decent quality files...
ETA is almost impossible to compute precisely. I wanted to omit it altogether, but decided to leave it. Now I am thinking it was a mistake. The reason is, that if the maps have a lot of "holes" as in empty tiles or missing levels, there is no recompression necessary, thus it speeds up. At the moment the ETA is a very dumb; "How many tiles have I done in the past x seconds, thus I'd need y seconds more to do all of them"... I might just do a pre-run phase to do a pre-analysis on ETA as well, but that might take a few seconds, which will come from the analysis time... I wasn't sure if it was worth it... Based on your feedback, it surely is now.
I hear you on the missing tile download! As far as I know GMDL already has that feature...
GPSFS_util is a simple post-processing program with some extra features. The real hard-work is in GMDL!