uuid: Add Universally Unique Identifier (UUID) utilities #77884

sorru94 · 2024-09-02T13:09:06Z

Add a basic implementation of UUID generating and manipulating functions.

jukkar · 2024-09-02T13:36:17Z

There are lot of uuid handling code in Bluetooth, should BT use this library?

sorru94 · 2024-09-02T14:47:49Z

There are lot of uuid handling code in Bluetooth, should BT use this library?

I'm unfamiliar regarding the Bluetooth use of the UUID. I see it uses 16 and 32 bits partial UUIDs, which are not defined in RFC 9562.
We could add a couple of defines similar to the ones in the bluetooth driver (BT_UUID_INIT_128 for example) to ease the transition.

jhedberg · 2024-09-02T20:22:41Z

There are lot of uuid handling code in Bluetooth, should BT use this library?

I'm unfamiliar regarding the Bluetooth use of the UUID. I see it uses 16 and 32 bits partial UUIDs, which are not defined in RFC 9562. We could add a couple of defines similar to the ones in the bluetooth driver (BT_UUID_INIT_128 for example) to ease the transition.

AFAIK at least the 16- and 32-bit Bluetooth UUIDs don't follow any non-Bluetooth standard. The way the Bluetooth core specification deals with these, is that in practice everything is a 128-bit UUID, however UUIDs which are derived from the so-called Bluetooth Base UUID can be compressed into 16- and 32-bit formats (and these again expanded back into 128 bit format). I have a feeling we'll need to keep the Bluetooth UUID API separate, but some proper research into their compatibility with these RFC specs needs to be done. We might at most e.g. provide conversion functions to convert back and forth between bt_uuid_t and uuid_t.

jukkar · 2024-09-03T05:08:34Z

@sorru94 could you check if there would be any non-BT code that could be changed to use this library?
At least in dhcpv6

zephyr/subsys/net/lib/dhcpv6/dhcpv6_internal.h

Line 83 in 0237d37

struct dhcpv6_duid_uuid {

there is uuid handling.

include/zephyr/sys/uuid.h

lib/uuid/uuid.c

tests/lib/uuid/src/main.c

andyross

Seems like there's an endianness collision between the existing code and the new library? UUIDs basically made a huge mess of this with that text format, and almost nothing does it "correctly" (if there even is a "correct"). But 100% for sure we should be doing it consistently, and I don't think we are?

include/zephyr/sys/uuid.h

andyross · 2024-09-05T00:04:35Z

tests/lib/uuid/src/main.c

+	uuid_t third_uuid_v4;
+	uuid_t first_uuid_v5;
+
+	uint8_t expected_first_uuid_v4_byte_array[16u] = {0x44, 0xb3, 0x5f, 0x73, 0xcf, 0xbd,


More drive-by: this is clearly byte-packing a big endian UUID. The first element is a 32 bit dword and should be stored "73, 5f, b3, 44" on little endian systems.

sorru94 · 2024-09-06T07:56:44Z

Seems like there's an endianness collision between the existing code and the new library? UUIDs basically made a huge mess of this with that text format, and almost nothing does it "correctly" (if there even is a "correct"). But 100% for sure we should be doing it consistently, and I don't think we are?

I'm unsure about the endianness for the UUID used/expected by the BT module (or other modules).
However, I tried to follow as much as possible RFC 9562 for this library where they state:

In the absence of explicit application or presentation protocol specification to the contrary, each field is encoded with the most significant byte first (known as "network byte order").

And:

Saving UUIDs to binary format is done by sequencing all fields in big-endian format. However, there is a known caveat that Microsoft's Component Object Model (COM) GUIDs leverage little-endian when saving GUIDs. The discussion of this (see [MS_COM_GUID]) is outside the scope of this specification.

I'm guessing the UUID in the Bluetooth module uses the COM GUIDs format where the first 4 bytes are represented using little endian while the last 6 are represented using big-endian.

I don't know if including a non-standard representation of UUIDs would be a good choice for this library. We could always introduce helper functions for conversion to the "COM GUIDs" format. Here or directly in the BT module.

What do you think? @andyross

include/zephyr/sys/uuid.h

lib/uuid/Kconfig

lib/uuid/uuid.c

andyross · 2024-09-06T23:33:54Z

However, I tried to follow as much as possible RFC 9562 for this library

Yeah... the problem is that "UUIDs" in practical code don't really and haven't ever followed that RFC, which is a post-hoc kind of cleanup thing. Windows GUIDs are the obvious example of something that is always converted LE, and the code @Thalley posted above certainly implies the BT inherited the convention.

I'm don't have enough of a stake here to demand one or the other, I'm just saying (1) whatever code we commit needs to be extremely clear about the endianness convention and (2) it should match the conventions used by any existing in-tree users (or else support both, I guess, but... yuck)

Thalley

The implementation looks OK to me. Unsure how and if we can use this in BT due to endianess, so I suggest to make it experimental so that we can modify things until we figure that part out.

include/zephyr/sys/uuid.h

Thalley · 2024-09-09T13:03:38Z

include/zephyr/sys/uuid.h

+ * An UUID in the standard big-endian RFC9562 representation `00112233-4455-6677-8899-AABBCCDDEEFF`
+ * has the equivalent little-endian COM GUID representation `33221100-5544-7766-8899-AABBCCDDEEFF`.


Is this right? The last 2 values, 8899 and AABBCCDDEEFF, don't need to be swapped?

The 8899 should be 9988 while the last one is correct. See the other related comment for more info.

What other related comment? Can you share the link?

Yes, if we follow the COM GUID representation the last two should not be swapped.

Sorry, I deleted the comment as It was my mistake. I replied on the other comment and will push the changes shortly.

lib/uuid/uuid.c

Thalley · 2024-09-09T13:06:07Z

lib/uuid/uuid.c

+	memcpy_swap(&out[first_part_start], &data[first_part_start], first_part_size);
+	memcpy_swap(&out[second_part_start], &data[second_part_start], second_part_size);
+	memcpy_swap(&out[third_part_start], &data[third_part_start], third_part_size);
+	memcpy_swap(&out[fourth_part_start], &data[fourth_part_start], fourth_part_size);
+	memcpy(&out[fifth_part_start], &data[fifth_part_start], fifth_part_size);


In the documentation for this function, only the first 3 parts are swapped, but this swaps the first 4. Why?

And why isn't the 5th swapped? (not familiar with the UUID spec)

The issue with the endianness is that the UUID spec defines UUID only as big-endian.
The interpretation of little-endian UUID is not universal.
From my understanding here are our options for the UUID:

Standard big-endian representation: 00112233-4455-6677-8899-AABBCCDDEEFF

little-endian representation inverting each block: 33221100-5544-7766-9988-FFEEDDCCBBAA

little-endian representation inverting the full UUID: FFEEDDCC-BBAA-9988-7766-554433221100

little-endian representation for Microsoft COM GUIs representation: 33221100-5544-7766-8899-AABBCCDDEEFF

I'm not sure which of those little-endian notations is used in the Bluetooth module.
However, from my very superficial research, I think the last one is the more widely adopted. It's also the only one referenced in the standard.

I'll need to read up on the UUID spec. Endianess is always a bit tricky, but it usually comes down the the language and terminology used in the specs and whether the blocks are values or arrays of values.

The blocks change based on the UUID version. For example, v5 is defined as follows:

0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | sha1_high | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | sha1_high | ver | sha1_mid | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |var| sha1_low | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | sha1_low | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Where sha1_* are all components of the same SHA-1 hash.
While UUID v1 is defined as follows:

0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | time_low | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | time_mid | ver | time_high | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |var| clock_seq | node | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | node | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

As a consequence, it might be difficult to find a generic conversion method.

lib/uuid/Kconfig

Thalley · 2024-09-12T10:01:48Z

So after the RFC and looking more into this, I've noticed the following:

4122, which is obsoleted by 9562, does seemingly not define any specific endianess
9562 defines UUIDs to be stored as big-endian
MS documentation seemingly only refers to 4122
Haven't found any references to 4122 or 9562 from the BT specs
If we want to be 9562 compliant and treat all UUIDs a big-endian, then it's pretty each to support
If we want to support converting between little and big endian, then it's more tricky. Each version of UUID (version 1 to 8) have slightly different structures, and ideally each value in each version should be converted. As far as I can tell 9562 does not define how to represent those as strings, because it is big-endian.
MS converts the first 3 values to little endian (https://devblogs.microsoft.com/oldnewthing/20220928-00/?p=107221) and then treats the 6-byte MAC address as a 6-octet array instead of a 6-octet value. MS seemingly also treats time_high and ver as a single value, instead of 12 and 4 bit values respectively. MS seemingly uses UUID version 1 (https://www.rfc-editor.org/rfc/rfc9562.html#name-uuid-version-1) for their GUIDs (UUIDs). The way that MS treats UUIDs is a complete mess 🙃

To continue with this PR, I think we need to ask what it is we want to achieve. Do we want to library that implements 9562? Do we want a library that is compatible with MS COM? Do we want a library that is compatible with Bluetooth?

The 9562 RFC has this nice little paragraph in the beginning that pretty much sums up the mess:

The use of UUIDs is extremely pervasive in computing. They comprise the core identifier infrastructure for many operating systems such as Microsoft Windows and applications such as the Mozilla Web browser; in many cases, they can become exposed in many non-standard ways.

The purpose of 9562 is to take RFC 4122 and properly standardize it.

Thalley

No additional comments from me.

It would have been nice to have a defined use case for this implementation (e.g. a sample or something), but not sure it's necessary. It's somewhat odd to add support for something without any uses, especially since this cannot be used with e.g. Bluetooth, but maybe the SOF module or something like that could use this in the future.

jhedberg · 2024-09-20T11:17:10Z

@sorru94 wouldn't it make sense to also include a uuid_copy(uuid_t dst, const uuid_t src) API, since the underlying type is an array instead of e.g. a struct?

sorru94 · 2024-09-20T14:23:12Z

@sorru94 wouldn't it make sense to also include a uuid_copy(uuid_t dst, const uuid_t src) API, since the underlying type is an array instead of e.g. a struct?

Yes, I think it would be helpful. I will add it.

sorru94 · 2024-09-20T14:27:46Z

No additional comments from me.

It would have been nice to have a defined use case for this implementation (e.g. a sample or something), but not sure it's necessary. It's somewhat odd to add support for something without any uses, especially since this cannot be used with e.g. Bluetooth, but maybe the SOF module or something like that could use this in the future.

I agree that it's odd. However, the UUID is already in use in many places within the project. I hope this library will entice people to standardize its use.

Thalley · 2024-09-20T15:10:49Z

I agree that it's odd. However, the UUID is already in use in many places within the project. I hope this library will entice people to standardize its use.

Which project are you referring to here? The SOF?

sorru94 · 2024-09-23T07:19:58Z

I agree that it's odd. However, the UUID is already in use in many places within the project. I hope this library will entice people to standardize its use.

Which project are you referring to here? The SOF?

I was referring to the Zephyr project.
As @jukkar mentioned above dhcpv6 uses some form of UUIDs. Furthermore, I could find some UUID in fs/ext2.h, usb/bos.h, and drivers/tee.h.
Hopefully, once this driver is available those modules could transition to using this one instead of ad hoc implementations.

jhedberg · 2024-09-23T09:04:34Z

The one thing that concerns me a bit with the way you've defined the uuid_t type, is that it hides away the distinction of passing by value vs passing by reference. Maybe that's a good thing, but it's often nice to be able to see this very clearly in API usage by having pointer types vs value types, and using & to pass by reference, etc. Since the type that uuid_t is encapsulating is an array you end up in practice always passing by reference, which could come as a surprise, e.g. if a function expects to be able to modify an input parameter without it having any effects on the caller. I don't have a strong opinion either way, and there does seem to be benefits of the current solution, but I'm wondering if you did some careful comparison & analysis of defining the type as:

typedef struct {
	uint8_t uuid[UUID_SIZE]
} uuid_t;

vs the current:

typedef uint8_t uuid_t[UUID_SIZE];

Thalley · 2024-09-23T09:13:56Z

The one thing that concerns me a bit with the way you've defined the uuid_t type, is that it hides away the distinction of passing by value vs passing by reference. Maybe that's a good thing, but it's often nice to be able to see this very clearly in API usage by having pointer types vs value types, and using & to pass by reference, etc. Since the type that uuid_t is encapsulating is an array you end up in practice always passing by reference, which could come as a surprise, e.g. if a function expects to be able to modify an input parameter without it having any effects on the caller. I don't have a strong opinion either way, and there does seem to be benefits of the current solution, but I'm wondering if you did some careful comparison & analysis of defining the type as:
typedef struct {
	uint8_t uuid[UUID_SIZE]
} uuid_t;
vs the current:
typedef uint8_t uuid_t[UUID_SIZE];

Doesn't our coding guidelines (or the Linux kernel guidelines) suggest not to typedef structs?
https://www.kernel.org/doc/html/v4.10/process/coding-style.html#typedefs

Another alternative would be to do

struct uuid {
    uint8_t uuid[UUID_SIZE]
};

and then pass around struct uuids which can be passed by value, by mutable reference or const reference (which I believe is the point of your suggestion here as well)

jhedberg · 2024-09-23T09:18:21Z

@Thalley my impression is that Zephyr is much more relaxed to the usage of typedefs, but there should always be a some good justification for their usage.

jukkar · 2024-09-24T04:37:49Z

I was referring to the Zephyr project.
As @jukkar mentioned above dhcpv6 uses some form of UUIDs. Furthermore, I could find some UUID in fs/ext2.h, usb/bos.h, and drivers/tee.h.
Hopefully, once this driver is available those modules could transition to using this one instead of ad hoc implementations.

I would suggest that at least one user of this api could be found and converted to use it in this pr. It is a bit weird to have an api in upstream that nobody uses.

Add UUID generation and parsing utilities compliant with RFC9562. Signed-off-by: Simone Orru <[email protected]>

Add some tests for the UUID utilities. Signed-off-by: Simone Orru <[email protected]>

Include the UUID utilities in the Miscellaneous documentation page. The UUID is being placed in a new Identifier APIs section. Signed-off-by: Simone Orru <[email protected]>

sorru94 · 2024-10-03T16:05:45Z

@Thalley and @jhedberg. As for your suggestion, I replaced the type definition with a struct. Let me know if it was what you intended.
@jukkar I tried to use this library in the ext2 library, just to have one that uses it.

Thalley · 2024-10-03T17:52:17Z

@sorru94 Looks like CI tests are failing, please have a look

Replace the typedefined array uuid_t with a struct named uuid. Signed-off-by: Simone Orru <[email protected]>

Use the sys/uuid.h library to generate and hold the ext2 UUID. This commit also includes some formatting for the ext2 library. Signed-off-by: Simone Orru <[email protected]>

Thalley · 2024-10-04T12:22:35Z

include/zephyr/fs/ext2.h

@@ -27,7 +29,7 @@ struct ext2_cfg {
 	uint32_t block_size;
 	uint32_t fs_size; /* Number of blocks that we want to take. */
 	uint32_t bytes_per_inode;
-	uint8_t uuid[16];
+	struct uuid uuid;


This is an API change, isn't it?

If it is a stable API, then it needs to follow https://docs.zephyrproject.org/latest/develop/api/api_lifecycle.html#introducing-breaking-api-changes

An alternative which is a bit dirty would be to do

Suggested change

struct uuid uuid;

union {

uint8_t uuid[16];

struct uuid uuid;

};

So that it's not breaking. I'll leave this up to the maintainer(s) of the fs

Thalley · 2024-10-04T12:24:59Z

include/zephyr/sys/uuid.h

+ * @retval 0 The UUID has been correctly copied in @p dst
+ * @retval -EINVAL @p dst is not acceptable
+ */
+int uuid_copy(struct uuid *dst, struct uuid src);


One thing I've noticed now is that you more often than not pass by value (e.g. struct uuid src) rather than by reference (struct uuid *dst).

Is that something you have considered? Since the struct is 16 octets, that's quite a lot of data being passed around compared to a pointer (typically 4 octets).

Should the API primarily use pass by reference instead of value?

Thalley · 2024-10-04T12:28:36Z

include/zephyr/sys/uuid.h

+ * @retval 0 The UUID has been correctly copied in @p dst
+ * @retval -EINVAL @p dst is not acceptable
+ */
+int uuid_copy(struct uuid *dst, struct uuid src);


In most functions you also have (src, dst) (e.g. uuid_from_buffer(const uint8_t data[UUID_SIZE], struct uuid *out); , but this functions does (dst, src).

Consider making them consistent

Thalley · 2024-10-04T12:31:17Z

subsys/fs/ext2/ext2_format.c

+	in->i_mode = sys_cpu_to_le16(EXT2_DEF_DIR_MODE);
+	in->i_uid = 0;
+	in->i_size = sys_cpu_to_le32(nblocks * cfg->block_size);
+	in->i_atime = 0;
+	in->i_ctime = 0;
+	in->i_mtime = 0;
+	in->i_dtime = 0;
+	in->i_gid = 0;
+	in->i_blocks = sys_cpu_to_le32(nblocks * cfg->block_size / 512);
+	in->i_flags = 0;
+	in->i_osd1 = 0;
 	in->i_generation = 0;
-	in->i_file_acl   = 0;
-	in->i_dir_acl    = 0;
-	in->i_faddr      = 0;
+	in->i_file_acl = 0;
+	in->i_dir_acl = 0;
+	in->i_faddr = 0;


Looks like you've misconfigured your editor to format all lines in the file, and not only the lines you've changed.

Please revert all format changes that are not changed

zephyrbot added the area: Base OS Base OS Library (lib/os) label Sep 2, 2024

zephyrbot requested review from andyross, dcpleung, nashif and peter-mitsis September 2, 2024 13:09

zephyrbot assigned andyross and nashif Sep 2, 2024

jukkar requested a review from jhedberg September 2, 2024 13:36

carlescufi requested a review from Thalley September 2, 2024 15:11

Thalley reviewed Sep 4, 2024

View reviewed changes

andyross reviewed Sep 5, 2024

View reviewed changes

sorru94 requested a review from andyross September 6, 2024 07:57

sorru94 force-pushed the add-uuid-utilities branch from 6941a58 to b42190f Compare September 6, 2024 09:04

sorru94 requested a review from Thalley September 6, 2024 09:12

Thalley reviewed Sep 6, 2024

View reviewed changes

sorru94 force-pushed the add-uuid-utilities branch from b42190f to 8ed2c71 Compare September 9, 2024 12:33

sorru94 requested a review from Thalley September 9, 2024 12:35

sorru94 force-pushed the add-uuid-utilities branch 3 times, most recently from 64065cc to 0ab6537 Compare September 9, 2024 12:54

Thalley reviewed Sep 9, 2024

View reviewed changes

sorru94 force-pushed the add-uuid-utilities branch from 0ab6537 to f7367b9 Compare September 11, 2024 10:46

sorru94 requested a review from Thalley September 11, 2024 10:46

Thalley previously approved these changes Sep 20, 2024

View reviewed changes

sorru94 dismissed Thalley’s stale review via 45c85d4 October 3, 2024 13:22

sorru94 force-pushed the add-uuid-utilities branch 2 times, most recently from c78c2df to 3b99fa7 Compare October 3, 2024 14:43

sorru94 added 3 commits October 3, 2024 17:58

uuid: Add UUID utilities

7decdbe

Add UUID generation and parsing utilities compliant with RFC9562. Signed-off-by: Simone Orru <[email protected]>

uuid: Add tests for UUID utilities

18c480a

Add some tests for the UUID utilities. Signed-off-by: Simone Orru <[email protected]>

uuid: Add documentation for UUID utilities

578f70b

Include the UUID utilities in the Miscellaneous documentation page. The UUID is being placed in a new Identifier APIs section. Signed-off-by: Simone Orru <[email protected]>

sorru94 force-pushed the add-uuid-utilities branch from 3b99fa7 to 214931f Compare October 3, 2024 15:59

zephyrbot added the area: File System label Oct 3, 2024

zephyrbot requested review from de-nordic and Laczen October 3, 2024 15:59

sorru94 force-pushed the add-uuid-utilities branch from 214931f to 318f323 Compare October 3, 2024 16:02

sorru94 force-pushed the add-uuid-utilities branch 2 times, most recently from 2ccc2cd to 6e5b0fe Compare October 4, 2024 09:27

sorru94 added 2 commits October 4, 2024 12:04

uuid: Replace typedefined array with struct

b0189fd

Replace the typedefined array uuid_t with a struct named uuid. Signed-off-by: Simone Orru <[email protected]>

fs: ext2: Use the UUID library

668ca04

Use the sys/uuid.h library to generate and hold the ext2 UUID. This commit also includes some formatting for the ext2 library. Signed-off-by: Simone Orru <[email protected]>

sorru94 force-pushed the add-uuid-utilities branch from 6e5b0fe to 668ca04 Compare October 4, 2024 10:05

sorru94 requested a review from Thalley October 4, 2024 12:00

Thalley requested changes Oct 4, 2024

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

uuid: Add Universally Unique Identifier (UUID) utilities #77884

uuid: Add Universally Unique Identifier (UUID) utilities #77884

sorru94 commented Sep 2, 2024

jukkar commented Sep 2, 2024 •

edited

Loading

sorru94 commented Sep 2, 2024 •

edited

Loading

jhedberg commented Sep 2, 2024

jukkar commented Sep 3, 2024

andyross left a comment

andyross Sep 5, 2024

sorru94 commented Sep 6, 2024 •

edited

Loading

andyross commented Sep 6, 2024

Thalley left a comment

Thalley Sep 9, 2024

Thalley Sep 11, 2024

sorru94 Sep 11, 2024

sorru94 Sep 11, 2024

Thalley Sep 9, 2024

sorru94 Sep 11, 2024 •

edited

Loading

Thalley Sep 11, 2024

sorru94 Sep 11, 2024

Thalley commented Sep 12, 2024

Thalley left a comment

jhedberg commented Sep 20, 2024

sorru94 commented Sep 20, 2024

sorru94 commented Sep 20, 2024

Thalley commented Sep 20, 2024

sorru94 commented Sep 23, 2024

jhedberg commented Sep 23, 2024

Thalley commented Sep 23, 2024

jhedberg commented Sep 23, 2024

jukkar commented Sep 24, 2024

sorru94 commented Oct 3, 2024

Thalley commented Oct 3, 2024

Thalley Oct 4, 2024

Thalley Oct 4, 2024

Thalley Oct 4, 2024

Thalley Oct 4, 2024

		* An UUID in the standard big-endian RFC9562 representation `00112233-4455-6677-8899-AABBCCDDEEFF`
		* has the equivalent little-endian COM GUID representation `33221100-5544-7766-8899-AABBCCDDEEFF`.

uuid: Add Universally Unique Identifier (UUID) utilities #77884

Are you sure you want to change the base?

uuid: Add Universally Unique Identifier (UUID) utilities #77884

Conversation

sorru94 commented Sep 2, 2024

jukkar commented Sep 2, 2024 • edited Loading

sorru94 commented Sep 2, 2024 • edited Loading

jhedberg commented Sep 2, 2024

jukkar commented Sep 3, 2024

andyross left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sorru94 commented Sep 6, 2024 • edited Loading

andyross commented Sep 6, 2024

Thalley left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sorru94 Sep 11, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Thalley commented Sep 12, 2024

Thalley left a comment

Choose a reason for hiding this comment

jhedberg commented Sep 20, 2024

sorru94 commented Sep 20, 2024

sorru94 commented Sep 20, 2024

Thalley commented Sep 20, 2024

sorru94 commented Sep 23, 2024

jhedberg commented Sep 23, 2024

Thalley commented Sep 23, 2024

jhedberg commented Sep 23, 2024

jukkar commented Sep 24, 2024

sorru94 commented Oct 3, 2024

Thalley commented Oct 3, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jukkar commented Sep 2, 2024 •

edited

Loading

sorru94 commented Sep 2, 2024 •

edited

Loading

sorru94 commented Sep 6, 2024 •

edited

Loading

sorru94 Sep 11, 2024 •

edited

Loading