New Config Files (introduced in v3.20)

Synchronet v3.20 introduces a new primary configuration file format using ini file syntax. Sysops can upgrade a Synchronet v3.1x configuration to v3.20 by running 'jsexec upgrade_to_v320.js' either before (preferably) or after upgrading their Synchronet executable binary files to v3.20. The update.js included with Synchronet v3.20 (invoked with 'jsexec update') will automatically invoke the 'upgrade_to_v320.js' as needed.

The existing ctrl/*.cnf files are converted one-to-one to .ini equivalents:

• ctrl/main.cnf -> main.ini
• ctrl/msgs.cnf -> msgs.ini
• ctrl/file.cnf -> file.ini
• ctrl/xtrn.cnf -> xtrn.ini
• ctrl/chat.cnf -> chat.ini
• node*/node.cnf -> node.ini

With the exception of a few never-used or no-longer-used configuration fields, the .cnf files should be completely represented by the newly created .ini files. Once converted, the old .cnf files will no longer be used and may be deleted if desired. If you're weary of the upgrade process, keep those old .cnf files around so that you can revert to Synchronet v3.1x if needed or even re-convert the .cnf files by executing upgrade_to_v320.js again if/when any issues in the script have been found and fixed later.

The now-legacy .cnf files were introduced in Synchronet v2.0 back in 1994. These .cnf files were designed for fast-loading: structured binary files with fixed length records comprised of fixed length fields (configuration properties) in a fixed order. Unfortunately, this lack of flexibility means it has also been an ongoing-challenge to add new configuration fields or extend the size of existing configuration fields. Most records included “padding” bytes for future expansion, but the future is now past and most of those bytes were consumed with new and expanded configuration fields; we've consumed the growth budget of most of the configuration records already. Now it'll be much easier to add new configuration fields and extend the size of existing fields when desired (e.g. no more 8-character internal code suffix limits!).

Additionally, it has been a challenge to keep track of changes to the stock configuration files stored in the Synchronet source code repository. Revision control systems (e.g. CVS, Git) can track changes in text files much better than changes to binary files. Moving to text-based configuration files remedies this change management issue and allows sysops to easily search and compare the contents of Synchronet configuration files (e.g. comparing with backed-up or stock/original versions of .ini files).

Yes, each .cnf file has now been replaced with an equivalent .ini file containing LF or CRLF-terminated lines of ASCII text conforming to the Synchronet supported .ini file syntax.

Interestingly, Synchronet v1 also used text configuration (*.cfg) files for the vast majority of its configuration settings, however the format was not as flexible as .ini and the large .cfg files were slow to load.

The flexible format of .ini files does mean that they take more compute (and time) to parse than fixed-structure (e.g. .cnf) files, but through better programming and the power of modern computers, the difference should be negligible.

Although the new .ini files can be directly edited using pretty much any plain-text editor, the primary supported method of manipulating these files is via the Synchronet Configuration Utility (SCFG). Any manually added comments (lines beginning with a semicolon) or white-space changes will be lost when the file is subsequently edited using SCFG. Unlike other Synchronet initialization files (e.g. sbbs.ini), there's been no effort made to make these new configuration files particularly easy to read (e.g. optional comments, named bits for bit-fields) and no effort made to maintain the exact key ordering or formatting used during any manual edits of the files. If a sysop chooses to make manual edits, they should either no longer use SCFG or be okay with the idea that the files will be completely re-formatted when subsequently edited with SCFG.

The .ini file format was chosen because this is already a first class data file format in Synchronet, supported equally well in the C/C++ code and JavaScript. Other structured text formats were considered (e.g. JSON, YAML), the pros and cons were weighed, and .ini won.

Thanks to mcmlxxix for his cnflib.js load library that made this configuration migration (the creation of upgrade_to_v320.js) relatively painless as Synchronet v3.20 maintains no knowledge of or ability to parse .cnf files directly.

• history index