Age | Commit message (Collapse) | Author |
|
This probably isn't the best fix, but it seems to make Facebook
connections stable for me.
After applying PR #497 you end up with frequent disconnects with:
```
Failed to read thrift: api.c:1515 fb_api_cb_publish_mst: assertion 'id == 2' failed
```
@.baltitenger said they checked that because that was the value in the
messages they saw, and they didn't know what other values would mean.
It seems messages with a different value don't break things (but we're
probably not handling them properly), but not crashing on them is
certainly better than crashing on them.
Fixes #496
Signed-off-by: Phil Dibowitz <phil@ipom.com>
|
|
Based on bitlbee/bitlbee-facebook#208
Fixes: #498
|
|
Fix taNewMessage bug
|
|
based on nuxi's PR on bitblee-facebook
|
|
|
|
- win32-build: fix version string format
- obs: put the ~ back in the version string for rpm packages
|
|
|
|
|
|
Forgot some things.
I don't understand m4 so let's do the obvious/dumb thing instead.
|
|
See bugs #449 and #410 respectively
This commit changes version string formatting a bit, now the
obs-generated packages will mention this version string too.
The version string obs.sh generates are stupid, we're going to need to
prefix things with an epoch like '1:' probably, assuming distros support
that anyway.
|
|
|
|
|
|
|
|
|
|
|
|
Thanks thorgrin for pointing this out
|
|
Thanks to mandree for noticing it's EAGAIN
See bug 410 for details
|
|
See also c22b43bb. Turns out mqtt.c had the same bug. See bug 403.
|
|
|
|
https://github.com/bitlbee/bitlbee-facebook/commit/99e31624bf9e88b3002e05514db904d5aad35db6
This workflow is terrible
Not making a release for this one, it affects a tiny amount of users who
definitely know how to compile stuff.
|
|
I like empty commits
|
|
This isn't the fix. This is so that people see a recent commit that says
something is fixed and stop asking where the fix is, because they think
a commit from two weeks ago can't possibly be the fix. But it is.
Handling support for these cases is my least favorite part of this thing
|
|
Sigh. Pretend this is 0.9.5.1 or whatever.
|
|
|
|
|
|
Patch 04-revert-gio.patch now also reverts upstream commit
38402:0f253504b9dd ("facebook: Fix reading MQTT ping responses")
which is a gio specific fix
It also partially reverts 38433:361c801c4536 ("Remove purple_timeout_*
function usage") in one place where it conflicted
Those were the actual conflict fixes, the rest are just refreshing line
numbers to apply cleanly
|
|
* Fix "Failed to read thrift" with unknown fields in /t_p payload
Fixes #357
|
|
|
|
Changes previously included in this repository as patches:
* Send user agent strings in all HTTP requests and in MQTT connect
* Add new ThreadListQuery-like hash, for initial seq id only
* Add content-type header to all requests, fixes http error 400
* Change error message when sync_sequence_id is missing, for clarity
Changes ported from bitlbee-facebook:
* Don't fail on groupchats with only two members including self
Those are created by the "Marketplace" feature
* Use FetchContactsDeltaQuery for contact sync
This has a number of benefits:
- Most of the time the contact sync reply will be empty
- We can do contact sync more frequently (It's 5 mins now, was 30)
- Figuring out what contacts were added or removed is much simpler and
less likely to get things wrong.
- Non-friends are no longer accidentally removed because there's no need
to compare contact lists
- On accounts with lots of friends this gets rid of one source of CPU
usage spikes
- Less load for facebook's servers (lol)
* Don't disconnect when trying to join a chat after being kicked
* Extract target url from fbrpc:// URLs
* Handle new style topic/groupchat membership events
* Fix setting channel topics (tid=t_id.1234 is no longer valid)
* Fix creating threads, inviting, kicking members (misc api changes)
Changes specific to purple-facebook:
* Add everyone to "Facebook Friends" group and persist Non-Friends
* Don't save access token if not remembering password
* Save names of non-friends in the local alias in blist.xml
* Add "Mark messages as read only when available" setting
|
|
|
|
|
|
|
|
|
|
|
|
It was getting hard to distinguish between users who never managed to
get the latest version of the plugin and users who are getting the error
even with the newest version.
|
|
The content-type header became required for /method/auth.login and
other endpoints with POST data like image attachments.
|
|
|
|
Fixes "Failed to parse thread information" errors when joining channels.
Facebook rolled a hotfix update today and they are requiring the same
user agent for HTTP requests too. As far as I can see the main affected
request is ThreadQuery.
Hopefully this also solves issues some a tiny minority of people were
still having with sync sequence ids.
|
|
|
|
Since it's used for different things and parameters are all wrong with
the newer query hash, but the seq id query doesn't use parameters.
|
|
Looks like this is what they use to discontinue older versions.
An empty user agent is treated as too old.
This fixes following error when sending messages:
This app version no longer supports messaging. To continue to send
and receive messages, please update your app to a newer version or
use the mobile web version of Facebook.
|
|
|
|
This hash represents a graphql query cached in the server. We were using
an old version of the hash that broke for some reason, possibly related
to the fact that fb messenger versions older than 5 are being
discontinued. For reference, the oldest parts of the plugin were written
when 4 was the latest version, most parts are updated to 33 or 38, and
the newest parts are based on 100 or so.
Thanks EionRobb for coming up with this fix so quickly.
|
|
|
|
It installs the unused/missing GTE_CyberTrust_Global_Root CA.
|
|
Shot in the dark to fix checkpoint issues, but had to be done anyway.
|
|
Relevant changes:
https://bitbucket.org/pidgin/main/pull-requests/167
* fb_api_cb_contacts: Make all the other fields optional too
* fb_http_urlcmp: more loose comparison, to avoid showing urls twice
* Prevent disconnections on 509 errors, "Invalid attachment id"
* Make $.hugePictureUrl.uri optional
* Store sent message id in lastmid, to deduplicate echoed messages
|
|
|
|
|
|
This reverts commit fe068747dfd1085702458ba3182abdbe50ee4c06.
Okay this was dumb. Turns out there's one spec at dist/ too.
|