Sopel 6.6.0 has been a long time coming, but it contains a metric ton of fixes and enhancements. The big stuff is yet to come in Sopel 7, though; stay tuned! (You can get some hints by browsing the GitHub milestones.)
View and contribute toward Sopel’s budgetary needs by visiting the project’s new Open Collective page.
Now that Sopel 6.6.0 is released, the focus will shift from preparing the new release to migrating the website and documentation off of Embolalia’s personal infrastructure. Once this migration is complete, Sopel’s documentation will be updated to reflect the changes/additions listed below. Questions about the new (and old) API methods are welcome in #sopel on Freenode, of course, as always. Thank you for bearing with this transitional period!
Importantly, a couple of broken modules have been removed in this release. For easy reference, they and their PyPI-installable replacements are listed near the top of the “Module changes” section.
Because this is the last release planned before Sopel 7, it’s important to note some upcoming changes. They’re mostly for module developers, or for bot owners who run unmaintained third-party modules (which might need manual updating).
But first, a user-facing change. Sopel is drifting toward a more streamlined release model, which will see most of the core module library split out into separate packages. Moving modules to their own packages will enable faster fixes when APIs change, and easier module adoption by members of the community.
Further process details will appear in future changelogs as modules are moved out of core, but the plan is to have Sopel “require” modules it used to include for a release or two after their removal. Existing installations should thereby have a smooth transition, as long as they get updated reasonably often. To discuss this plan, visit issue #1291.
And now, for the developer stuff.
There will be some attention given to how
Identifier objects work in Sopel 7.
In particular, the way lowercase works appears to have been wrong for some time.
Subscribe to issue #1389 for updates, or to join the discussion on the best
plan for addressing this.
Also in Sopel 7, the following deprecated functions in
sopel.web will be
get_urllib_object(). Module authors
should replace these with equivalent code using
requests. If you don’t know
what that is, you’re in for a treat: https://pypi.org/project/requests/
sopel.web functions might also move in a future major release. See
#1318 and comment your opinion if you are a module developer. There is still
plenty of time to give feedback; nothing will happen before Sopel 8 at the
Support for multiple help examples on each module command is planned for Sopel 7 and will allow for clearer command and argument documentation. The current help system uses a deterministic but unintuitive method to choose a single example to output. The current plan is to make the new multi-example behavior opt-in, so only help for modules that are explicitly updated to support it will change. Follow this initiative in #1200.
Bot owners and admins might appreciate the new ignore system proposal, which would (among other things) make “hostmask” blocks actually block hostmasks! See #1355 to participate in the design process; this change is tentatively planned for Sopel 7 but may be pushed back.
And finally, for users who hate SQLite with a passion: Work on bringing back alternate DB support has begun! It’s unclear how soon this will be ready, but users who want to switch off SQLite to another database engine might be able to do so as early as Sopel 7. Follow the ups and downs of this long-awaited journey at #1446 for the time being.
- Added emoticons module: includes
.lenny, and other common kaomoji
- Added instagram module: displays information about Instagram links in chat
- Removed movie: The OMDb API went private some time ago (Q2 2017)
sopel-modules.imdbon PyPI for a functional replacement
- Removed weather: Yahoo deprecated its weather service, the last known keyless weather API
sopel-modules.weatheron PyPI for a functional replacement that also adds the forecast command originally planned to launch with Sopel 6.6.0.
- admin module’s
.setcommand handles spaces properly now
- bugzilla no longer spits out an error on shutdown if its domain list is empty
- dice module’s
.dcommand gives friendlier errors, and its
.chcommand has cleaner output (will be further improved in a later release)
- etymology module updated to work with etymonline.com’s changes
- ip module can now look up nicknames (user’s hostname) & IPv6 addresses
- ip module switched from deprecated GeoIP database to GeoIP2
- isup module can be forced to ignore untrusted/broken HTTPS with a new command
- lmgtfy module gained improved output URLs (HTTPS, encoded query) and help examples
- reddit module now picks up links under all commonly used subdomains
- reload’s habit of duplicating packaged modules’ command output has been
- There are still some edge cases (like reloading a module with a removed function) that might still cause problems, but the majority of module updates should be OK now.
- This feature will continue to improve over the next few releases.
- search module should behave better when encountering Unicode in URLs on py2
- search now warns
.gusers of a bug affecting multiple “site:” operators in a single query
- This is a bug in DuckDuckGo’s plain-HTML SERP, and they have ignored all attempts to report it. Sopel’s DuckDuckGo code might be rewritten to use an unaffected interface in the future.
- tld module gracefully handles missing arguments now
- translate now handles API failures gracefully
- url module can now create a short URL for convenience when fetching title
- only URLs longer than configured minimum length
- off by default
- url module will say an error if
- url module’s URL finder will ignore suspicious trailing punctuation, making the auto-title feature more reliable when links are included in sentences
- version module no longer prints “wat” to the console/log when answering CTCP
- wiktionary module now decodes HTML entities in its output before sending, and it should also now give correct results for affixes & definitions containing reference tags
- xkcd fetching by comic title should work again (switched back-end search provider)
Many modules updated to use HTTPS in both API calls and output
- Added default value of
- Lets Sopel respond to (a) nickname(s) other than the name it uses for its IRC nick
- Made handling
CNAME‘d TLS hostnames more forgiving
- Raw logging is no longer enabled by default
- Comparisons between Identifiers and other types were improved
- Sopel no longer sends the sticky modifier for capabilities
- Database tables are now created using “IF NOT EXISTS” to reduce error potential
sopel.webcalls in various modules updated to use
- Intents/CTCP handling fixed
- Fixed more cases where module shutdown routines weren’t running
- New formatting methods:
- With the exception of italics, these are all “niche” formatting codes that have relatively limited client support, but they are included for completeness.
sopel.db.get_uri()was returning an incorrectly formatted result, now fixed
- Fixed populating
bot.channelsprivileges dicts on connect
- Fixed updating
Userobjects when receiving
NICKevents from the server
- Documentation added/corrected in several places
- Command prefix in examples is now assumed to match the default regex pattern (currently
- This fixes some issues with incorrect help output, but might cause new problems in modules that use non-standard prefixes in their examples.
- Sopel’s maintenance team regrets not providing advance warning of this change, but the potential impact is limited to perhaps a few misplaced characters in help output, at most—same as the bug itself, but likely with a much lower incidence.