| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| | |
|
| |
|
|
| |
C/C++ time handling is awful, pass it on
|
| | |
|
| |\ |
|
| | | |
|
| | | |
|
| | | |
|
| | |\ |
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This update fixes the underlying issue behind an infinite-stalling bug
that has been around since differencebot was written. The four bots that
download images from ImageNet (difference, advice, grunge, capital) had
the same small logic error whereby the catch block, written to handle the
situation when an image had failed to download, incorrectly took the
exception by value rather than by const reference. While inefficient,
this should not have caused any problems, but curlcpp, which the four
bots all use through libtwitter++'s dependence on it, had a bug in the
copy constructor of the curl_exception object which could cause a
deadlock situation. This went unnoticed for a while, despite the four
bots having serious downtime issues. When this was discovered, the bots
were patched so that the relevant catch block took the exception by
const reference. While this fixed things, it is also prudent to update
curlcpp to the newest version in which this bug has been fixed.
refs JosephP91/curlcpp#100
|
| | |/
| |
| |
| |
| |
| |
| |
| | |
While uploading media worked in development, as well as on production
when compiled in debug mode, it failed with memory corruption issues on
production in release mode. This turned out to be because curlcpp was
taking lvalue references to temporary std::string objects converted from
string constants, instead of just copying the string pointer.
|
| | | |
|
| | | |
|
| |/
|
|
| |
Because the streaming API will sunset on August 16th, it is essential to implement timeline polling so that the library can still be used to access tweets. This commit breaks the current API even for clients still using the streaming API because the OAuth connection data was pulled from twitter::client into twitter::auth. Other than that, almost everything works the same, and if a client wants to update their libtwitter++ within the next week and a half, they are free to.
|
| |
|
|
| |
Notifications are now also mutable. Users and tweets no longer have helper methods for interacting with the client. Fixed a bug (possibly introduced by a change to the Twitter API) that caused non-reply tweets to be marked as unknown notifications.
|
| |
|
|
| |
The submodule contained around 73MB of benchmarks and tests that are not necessary for inclusion in this project. Thus, the submodule has been removed, and the 2.0.7 release of nlohmann/json has been added to the repository.
|
| | |
|
| | |
|
| | |
|
| |\ |
|
| | | |
|
| | |\ |
|
| | | | |
|
| | |/
|/| |
|
| | | |
|
| |/ |
|
| | |
|
| | |
|
| |
|
|
| |
Because a retweet is a tweet referenced from another tweet, the current implementation dynamically allocates memory for the retweeted status and has to manually copy the pointed to data around. It's not a very pretty implementation and may get tweaked at some point soon. I don't like having to implement the big five.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
You can now start a user stream and end it yourself. If it disconnects abnormally, it will reconnect with a backoff as described by Twitter. Some data structures have some fields parsed now; tweets have IDs, text, and authors. Users have IDs, screen names, and names. Notifications from the stream are parsed completely. The ability to follow and unfollow users has also been added, as well as the ability to get a list of friends and followers, and to reply to a tweet.
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Images (static and animated) and videos have been tested. Currently all media uploads occur in one large chunk; support to break down chunks will be added later.
|
| |
|
