Releases: rupor-github/InpxCreator
Small fixes
Correcting the way --out-dir works: if passed to the program do not create inpx file in "data" subdirectory but rather in out-dir directly and make sure that no empty "data" subdirectory is left after processing is done
Linux port
Linux port has been completed and verified on up-to-date Arch64 and Synology DSM 6.x - synology_braswell_916+: resulting INPX files are identical to Windows builds. Linux binaries are statically built, so they should run on wide variety of Intel 64 bits Linux-es.
Code changed some - lib2inpx lost "follow-symlinks" and gained "out-dir" parameters. Return code from libget2 and libmerge changed to work for Posix shells. There is a new bash script fb2_flibusta.sh (which basically does the same thing as fb2_flibusta.ps1).
Process multiple paths in libmerge
This allows to keep daily updates in a separate directory (to avoid triggering unnecessary backup)
Maintenance
gcc 6.2, mysql 5.7.15, boost 1.62
Skip backup files in archives
Sometimes when correcting books it makes sense to keep original copy in archive for backup purposes. From now on all files with extention .org in archive will be skipped during processing.
Post-conversion fixes
Squashed a couple of minor bugs
Moving to platform independence
- Deleted libsplit from distribution - slow and outdated, use libmerge instead.
- A lot of code changes in lin2inpx, mostly invisible from outside - moving to libiconv for all string encoding related needs. It should be relatively trivial now to produce working binaries for Linux.
- User visible: replacing NON-BREAKING-SPACE with simple SPACE in all strings from database.
Maintenance release
- Do not remove last last archive when next one is ready
- Move to gcc 6.1, go 1.7, MySQL 5.7.14
- Change fb2_flibusta.ps1 to rebuild "inpx" only when new merge archive is available. Remove all but last 5 downloaded db dumps directories
Different updates download and merge logic
Switching to glide for go vendoring.
Removing libmerge --inplace mode: it will merge updates into temporary archive until it reaches proper size at which point old "last" archive is removed and temporary merge archive renamed (becoming new "last" archive"). Another temporary merge archive is created after that (if necessary). This allows MyHomeLib to keep referring to "last" archive until new one "grows" avoiding unnecessary re-indexation.
Changing libget2 and lib2inpx to support new file naming convention ignoring temporary merge archive.
Changing libget2 http get timeout processing - now http read timeouts are properly handled, avoiding connection re-opening.
Cleaning libget2 error reporting with errors package.
Sorting archives
Added archive sorting - on the network _findfirst returns files in random order
gcc 5.4.0, go 1.6.2, boost refreshed