Fix CI failures #394

jean-roland · 2024-04-03T09:45:35Z

Following #389, CI is functional again on this branch and it revealed many small issues that are fixed in this PR, along with a merge of the main branch to bring the aforementioned changes.

There is still an issue regarding fragmentation that seems to involve zenoh. Edit: solved

- add missing dots in docs

- replace _z_n_qos_unmake with _z_n_qos_unmake_public in subscription.c

…ings

… from zenoh-cpp

* Add Flipper platform * Fix warnings and other platform builds

…ority-in-sample Add support for qos settings in sample

…zers from c++)

…ializers

…qos-designated-initializers Fix/remove qos designated initializers

…enoh-c

…tform-api-sync-with-zenoh-c platform api synchronization with zenoh-c

* feat: add ethtype to reth endpoint config * feat: type renaming * fix: remove unused macro call * feat: remove static raweth config * feat: use defines for separators * feat: add mapping function * fix: add default raweth mapping * feat: add debug traces on raweth config parsing * fix: ethtype validity test * fix: strtok parsing calls * chore: clang-format * fix: include first mapping entry in lookup * fix: appease the tyran codacy

* Expose timeout option in z_get_options_t * Add default get timeout

* Remove reading from stdin, align example implementations * Update examples tests * Adjust z_sub_thr output * Fix z_get single query examples * Replace pthread uses with z_mutex and z_condvar * Add multi-thread feature condition to z_get examples * Update error message for features absence * Update z_get expected output in modularity test * Update sample count for freertos single-thread examples

examples/unix/c11/z_queryable.c

examples/unix/c11/z_sub.c

Mallets · 2024-04-05T10:02:06Z

examples/unix/c11/z_sub_st.c

@@ -37,7 +37,7 @@ int main(int argc, char **argv) {
    const char *mode = "client";
    char *clocator = NULL;
    char *llocator = NULL;
-    int n = -1;
+    int n = 2147483647;  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

examples/unix/c99/z_sub.c

examples/unix/c99/z_queryable.c

Mallets · 2024-04-05T10:04:41Z

examples/freertos_plus_tcp/z_pub_st.c

@@ -30,6 +30,7 @@

 #define KEYEXPR "demo/example/zenoh-pico-pub"
 #define VALUE "[FreeRTOS-Plus-TCP] Pub from Zenoh-Pico!"
+#define N 2147483647  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

With a signed integer ~0 will be interpreted as -1. As this is also an example thing, I suggest we fix this with the z_sleep later.

You are right. Technically it should be an unsigned int and not an int.

Mallets · 2024-04-05T10:04:51Z

examples/freertos_plus_tcp/z_sub_st.c

@@ -27,13 +27,17 @@
 #endif

 #define KEYEXPR "demo/example/**"
+#define N 2147483647  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

Mallets · 2024-04-05T10:04:59Z

examples/unix/c11/z_pub.c

@@ -30,7 +30,7 @@ int main(int argc, char **argv) {
    const char *mode = "client";
    char *clocator = NULL;
    char *llocator = NULL;
-    int n = 10;
+    int n = 2147483647;  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

Mallets · 2024-04-05T10:05:10Z

examples/unix/c11/z_pub_st.c

@@ -27,7 +27,7 @@ int main(int argc, char **argv) {
    const char *mode = "client";
    char *clocator = NULL;
    char *llocator = NULL;
-    int n = 10;
+    int n = 2147483647;  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

Mallets · 2024-04-05T10:05:45Z

examples/unix/c99/z_pub.c

@@ -26,7 +26,7 @@ int main(int argc, char **argv) {
    const char *mode = "client";
    char *clocator = NULL;
    char *llocator = NULL;
-    int n = 10;
+    int n = 2147483647;  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

Mallets · 2024-04-05T10:06:00Z

examples/unix/c99/z_pub_st.c

@@ -26,9 +26,10 @@ int main(int argc, char **argv) {
    const char *mode = "client";
    char *clocator = NULL;
    char *llocator = NULL;
+    int n = 2147483647;  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

Mallets · 2024-04-05T10:06:24Z

examples/windows/z_pub_st.c

@@ -18,6 +18,8 @@
 #include <stdlib.h>
 #include <zenoh-pico.h>

+#define N 2147483647  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

examples/windows/z_queryable.c

examples/windows/z_sub.c

Mallets · 2024-04-05T10:07:02Z

examples/windows/z_sub_st.c

@@ -18,13 +18,18 @@
 #include <stdlib.h>
 #include <zenoh-pico.h>

+#define N 2147483647  // max int value by default


It's safer to use int n = ~0; in such a way we always have the guarantee n is max(int) regardless the actual size of int.

examples/zephyr/z_queryable.c

examples/zephyr/z_sub.c

examples/espidf/z_queryable.c

Mallets · 2024-04-05T10:20:31Z

src/system/espidf/network.c

+    uart_wait_tx_done(sock->_serial, 1000);
+    uart_flush(sock->_serial);
+    uart_driver_delete(sock->_serial);
+    zp_free(sock->after_cobs);


Shouldn't be z_free?

Mallets · 2024-04-05T10:20:37Z

src/system/espidf/network.c

+    uart_flush(sock->_serial);
+    uart_driver_delete(sock->_serial);
+    zp_free(sock->after_cobs);
+    zp_free(sock->before_cobs);


Shouldn't be z_free?

Mallets

Some minor changes, everything else LGTM.

jean-roland · 2024-04-05T13:20:38Z

We noticed some issues with examples:

Some still use sleep instead of z_sleep
We should use an unsigned int set to ~0 for the default loop count
But this is not the scope of this PR so this shall be adressed by another PR

Mallets · 2024-04-05T13:24:25Z

We noticed some issues with examples:

Some still use sleep instead of z_sleep

We should use an unsigned int set to ~0 for the default loop count
But this is not the scope of this PR so this shall be adressed by another PR

Agree

DenisBiryukov91 and others added 30 commits February 15, 2024 17:52

add support for qos settings in sample

84b11ad

- silence cpp compiler regading compound literals and type conversions,

d356881

- add missing dots in docs

fix _z_qos_t definition and doc

cb554fd

fix z_qos_t definition and doc

f0545eb

- replace binary literals (non-core C) with hexadecimal ones

6e10617

- replace _z_n_qos_unmake with _z_n_qos_unmake_public in subscription.c

make _z_n_qos_unmake static inline to prevent linking errors

29fca01

fix format

bb65bc8

reduce qos size to 1 byte; use getters to extract individual qos sett…

39cfc82

…ings

fix _z_n_qos_make to no longer use compund literals to avoid warnings…

245aade

… from zenoh-cpp

fix z_qos_default() signature

790bd62

Add Flipper platform (eclipse-zenoh#351)

3bab290

* Add Flipper platform * Fix warnings and other platform builds

Merge pull request eclipse-zenoh#348 from DenisBiryukov91/feature/pri…

2ac04a2

…ority-in-sample Add support for qos settings in sample

move z_n_qos_t related functions into .c (to hide designated initiali…

b5c702b

…zers from c++)

z_sample_t docs update to include qos

8e37ab8

move qos functions back to header and remove usage of designated init…

cac8b3d

…ializers

format

54c291e

Merge pull request eclipse-zenoh#354 from DenisBiryukov91/fix/remove-…

a657a1d

…qos-designated-initializers Fix/remove qos designated initializers

fix: Bump sphinx doc build dep to 7.2.6 (eclipse-zenoh#361)

e3707c1

replac zp_ prefix for platform functions that are also available in z…

a16d84c

…enoh-c

fix format

5216d7a

replace zp_ prefix with z_ for zp_random_u64

b21b11d

put back previously removed zp_ functions as deprecated

1f45c9d

added missing new line at the end of the file

c51ffe2

typo fix

d9ed7eb

move deprecated platform header to a separate folder

1c3bfa0

Merge pull request eclipse-zenoh#365 from DenisBiryukov91/feature/pla…

9c2cf4c

…tform-api-sync-with-zenoh-c platform api synchronization with zenoh-c

Fix bug reading from ws (eclipse-zenoh#370)

9f7925b

Expose timeout option in z_get_options_t (eclipse-zenoh#375)

0d9f25b

* Expose timeout option in z_get_options_t * Add default get timeout

jean-roland mentioned this pull request Apr 5, 2024

Rework owned sample #391

Merged