When compiling on buildroot I get:
`src/tls.c:235:28: error: ‘ssize_t’ undeclared (first use in this
function); did you mean ‘size_t’? `
To have this POSIX type available we will need to include
`sys/types.h` or `unistd.h`.
Before this change the user had to provide a password callback, even if the
PKCS#12 encoded file has no or an empty password.
This changes the behavior, so we first try to open the file without a
password and only then ask the user to provide one.
This fixes#204.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
In order to be able to distinguish between user-created entries and
library-internal entries, we mark each entry who it belongs to.
This allows then later to let the user manage only the queue entries
that were created by them.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
The cache is stored per connection object and is cleared on
* entry of wrong password
* release of connection object
* successful connection
It can be configured that libstrophe retries the password entry in case
the user entered a wrong password.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
In order to be able to load password-protected key files a password
callback was added.
This also adds support for PKCS#12 containers instead of certificate+key.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
The registered callback function will be called before connect() is
called on the socket, allowing tweaking of much more than just keepalive
settings.
This deprecates xmpp_conn_set_keepalive().
Includes helper/example callback function to set default keepalive
parameters that some library users may find useful.
In order to do proper deprecation of those API's we re-introduce them
so they can be removed in the next release. This will then also lead to
an ABI bump.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
In case a pointer to itself was received, the resolver looped inifinitely
until the stack overflows.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
src/handler.c: In function 'handler_system_delete_all':
src/handler.c:706:25: warning: 'key2' may be used uninitialized in this function [-Wmaybe-uninitialized]
hash_add(conn->id_handlers, key2, head);
Instead of using the internal DRBG-based RNG, use the `getrandom()` system
call where available.
Keep the existing version as fall-back for other systems that don't provide
that system call.
... and for cases where auto-detection fails or the user simply wants to
use the internal RNG a new configure-switch is provided.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
... to provide an easier way to find long paths.
After looking at some code in profanity-im/profanity#1605 I thought this
could help to make libstrophe-user code easier to write (and read).
Initially it started off as `xmpp_stanza_get_child_by_xpath()` but then
I was annoyed when it came to handling all those potential corner cases
like e.g. escaping URL's in the ns ... so here we go with a
vararg approach :)
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
`hash_iter_next()` returns a pointer to an element of the data structure
which is persistent and not altered while calling `hash_iter_next()`.
Therefor we don't have to dup the string, but simply can remember the
pointer.
In the case where the element would be `hash_drop()`'ped which contains
the key, the key is not accessed after it has been free'd.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
Based on the differences to libmesode this functionality has been added.
It allows a library-user to set a callback for cases where the TLS stack
can't verify a received certificate and let the end-user decide what to
do.
examples/basic implements an example handler of said functionality.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
This adds a new API function xmpp_ctx_set_verbosity() to set increased
verbosity levels.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
While reading through the code this seemed logical to do.
I haven't seen this occuring in the wild!
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
This returns early in case
1. no log handler is set
2. the default log handler is set and the log level would result in nothing
being printed.
No need to snprintf() and whatnot if there's no consumer.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
Put sending of data into its own function which takes ownership of the
data and therefor minimizes the number of allocations and frees.
This also slightly improves error handling and printing of debug
information.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
