tap_user usability tracking #87

fish4terrisa-MSDSM · 2023-04-10T04:32:22Z

( Sorry for my poor english)
The QEMU has a feature to bind the guest`s port to the host\s port.Will RVVM support it later?It\s impossible to support x11 or SDL on android,so the vnc is the only choice.It`s important to make it possible to connect to the guest using the network port.

The text was updated successfully, but these errors were encountered:

LekKit · 2023-04-10T09:39:14Z

Yeah it will be there in future. The networking is WIP as you can see from README, my attention is pretty scrambled over multiple things.
Can we convert this issue into tap_user usability tracking issue?

P.S. Is there a way to draw a raw framebuffer on Android using native GUI APIs? I am going to add JNI support and then we can have native Android RVVM app without VNC shenanigans. A while ago I expected SDL to be usable on Android and only later realized it isn't...

LekKit · 2023-04-10T09:53:36Z

Current `tap_user` issues/TODOs:

No way to do 2-tuple lookup in TCP, because hashmap API is limited. This prevents us from doing 2-tuple reuse and therefore TCP port forwarding is broken.
~~No TCP buffering, it just expects the guest is able to receive all packets in time. Sometimes this results in dead connections, pacman failing to download packages, etc etc.~~
Inefficiend locking, it could be much more scalable and with less overhead
Basic firewall/traffic limiting

LekKit · 2023-04-10T10:48:49Z

Maybe it is worth to add VNC support into RVVM itself. There is such feature in QEMU (It acts like a VNC server), but I am unaware how complicated that might be (At least we have a nice networking lib).

Doing that via guest means is arguably not an ideal solution (We don't see the boot process, requires user intervention, what about non-Linux guests?), and we need a proper Android port anyways. So I guess JNI + android GUI is the best long term solution for this usecase.

fish4terrisa-MSDSM · 2023-04-10T11:51:38Z

Yeah it will be there in future. The networking is WIP as you can see from README, my attention is pretty scrambled over multiple things. Can we convert this issue into tap_user usability tracking issue?

P.S. Is there a way to draw a raw framebuffer on Android using native GUI APIs? I am going to add JNI support and then we can have native Android RVVM app without VNC shenanigans. A while ago I expected SDL to be usable on Android and only later realized it isn't...

Ok i will change the issue name.

P.S. I dont know much about the android api .(in fact,my project heavily depends on alpine-term.I dont found any android app can draw a raw fb,even the termux-x11 or xserver xsdl dont use it).Besides, adding gui support is not so important i think .Many people dont need the gui support on android,they may prefer terminal .And if the port forwarding and the gpu support is done ,the user can start a vnc by themselves.A unused vnc causes profermance decrease. If you want to build the own RVVM app ,editing on my project may be a good idea ,just delete the descriptions of archlinux and you will get one .(My project is still undone ,it will take me about 2 weeks to fully move from QEMU to RVVM .And if its done ,i will notice you in the discussions area)

fish4terrisa-MSDSM · 2023-04-11T13:21:02Z

Maybe it is worth to add VNC support into RVVM itself. There is such feature in QEMU (It acts like a VNC server), but I am unaware how complicated that might be (At least we have a nice networking lib).

Doing that via guest means is arguably not an ideal solution (We don't see the boot process, requires user intervention, what about non-Linux guests?), and we need a proper Android port anyways. So I guess JNI + android GUI is the best long term solution for this usecase.

Thats true ,I prefer the vnc method ,its also useful on other platforms.The Android GUI method is a bit complicated (termux-x11 might be a exmaple , it has developed for more than 3y but still have large amounts of bugs) ,but adding a vnc viewer in the app might be a good idea as the user can connect to the machine without using other apps.

fish4terrisa-MSDSM · 2023-04-11T13:49:01Z

A good vnc viewer project is avnc, its license is GPL3.You can insert some parts of it into the app.

LekKit · 2023-05-06T13:18:44Z

Commit 833a16e solves all previous TCP traffic issues and running out of UDP ports.
It also includes groundwork for TCP NAT (host->vm connections) but the way to lookup socket via 2-tuple is still in the works.

Please report if you still have any TCP stability issues.

LekKit · 2023-05-07T09:11:00Z

Figured there is an issue with keepalive ACKs not being properly handled. Fixed locally, will be upstreamed soon.
(This only matters if you have completely idle connections that live for hours)

LekKit · 2023-05-08T16:14:20Z

TCP port forwarding works. However, there is an issue when connecting directly to 127.0.0.1 from the host, the guest gets a connection from 127.0.0.1 as well and gets confused, so the connection is dropped. I don't know how to solve that correctly yet.
Will be upstreamed soon.

fish4terrisa-MSDSM · 2023-05-09T01:43:36Z

A bug occured on NetBSD:

[root@archlinux ~]# echo $LANG
C.UTF-8
[root@archlinux ~]# curl www.google.com
WARN: net_poll_add() failed!
curl: (1) Received HTTP/0.9 when not allowed
[root@archlinux ~]# curl https://www.google.com
curl: (35) OpenSSL/3.0.8: error:0A0000C6:SSL routines::packet length too long
[root@archlinux ~]#

and when run curl www.google.com --http0.9,it will output the index.html of google,but the connection won`t close before timeout,so i think both the openssl error and the curl error is because that tcp connections cannot close accurately on NetBSD.The ping runs correctly ,so icmp have no problems.
When runs curl ip.sb --http0.9 to get the ip , it outputed

[root@archlinux ~]# curl ip.sb
curl: (1) Received HTTP/0.9 when not allowed
[root@archlinux ~]# curl ip.sb --http0.9
TTP/1.1 200 OK
Date: Tue, 09 May 2023 01:38:16 GMT
Content-Type: text/plain
Content-Length: 13
Connection: keep-alive
Cache-Control: no-cache
CF-Cache-Status: DYNAMIC
Report-To: {"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v3?s=cgybpWVU5jI3iZjC8NPmm7lMwYhsUfoaNkW7zU5VFKGN%2BxbuCpdO9E1Z%2Fl%2BKK3Rz3gAVbsJx9Pm9q9B9LGOwd0Lrt4dE5hLwJK1y4pfyf3X3ut8XYFoV"}],"group":"cf-nel","max_age":604800}
NEL: {"success_fraction":0,"report_to":"cf-nel","max_age":604800}
Server: cloudflare
CF-RAY: 7c4638d53bc13087-SEA
alt-svc: h3=":443"; ma=86400, h3-29=":443"; ma=86400

205.166.94.4

The 'H' of "HTTP/1.1" disappeared ,so both the length of the packet and the head of the packets might be wrong(?)
Output of ifconfig

[root@archlinux ~]# ifconfig
enp0s1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 192.168.0.100  netmask 255.255.255.0  broadcast 192.168.0.255
        inet6 fe80::8c2e:4fff:fed9:45f  prefixlen 64  scopeid 0x20<link>
        ether 8e:2e:4f:d9:04:5f  txqueuelen 1000  (Ethernet)
        RX packets 45  bytes 10639 (10.3 KiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 79  bytes 6345 (6.1 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10<host>
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

The dhcp just worked well.I haven`t tested udp connections yet,and I also don`t know much about how to test it.
p.s. : now the VMA worked well on sdf.Great!

fish4terrisa-MSDSM · 2023-05-09T01:51:14Z

A bug occured on NetBSD:

[root@archlinux ~]# echo $LANG
C.UTF-8
[root@archlinux ~]# curl www.google.com
WARN: net_poll_add() failed!
curl: (1) Received HTTP/0.9 when not allowed
[root@archlinux ~]# curl https://www.google.com
curl: (35) OpenSSL/3.0.8: error:0A0000C6:SSL routines::packet length too long
[root@archlinux ~]#

and when run curl www.google.com --http0.9,it will output the index.html of google,but the connection wont close before timeout,so i think both the openssl error and the curl error is because that tcp connections cannot close accurately on NetBSD.The ping runs correctly ,so icmp have no problems. When runs curl ip.sb --http0.9` to get the ip , it outputed

[root@archlinux ~]# curl ip.sb
curl: (1) Received HTTP/0.9 when not allowed
[root@archlinux ~]# curl ip.sb --http0.9
TTP/1.1 200 OK
Date: Tue, 09 May 2023 01:38:16 GMT
Content-Type: text/plain
Content-Length: 13
Connection: keep-alive
Cache-Control: no-cache
CF-Cache-Status: DYNAMIC
Report-To: {"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v3?s=cgybpWVU5jI3iZjC8NPmm7lMwYhsUfoaNkW7zU5VFKGN%2BxbuCpdO9E1Z%2Fl%2BKK3Rz3gAVbsJx9Pm9q9B9LGOwd0Lrt4dE5hLwJK1y4pfyf3X3ut8XYFoV"}],"group":"cf-nel","max_age":604800}
NEL: {"success_fraction":0,"report_to":"cf-nel","max_age":604800}
Server: cloudflare
CF-RAY: 7c4638d53bc13087-SEA
alt-svc: h3=":443"; ma=86400, h3-29=":443"; ma=86400

205.166.94.4

The 'H' of "HTTP/1.1" disappeared ,so both the length of the packet and the head of the packets might be wrong(?) Output of ifconfig

[root@archlinux ~]# ifconfig
enp0s1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 192.168.0.100  netmask 255.255.255.0  broadcast 192.168.0.255
        inet6 fe80::8c2e:4fff:fed9:45f  prefixlen 64  scopeid 0x20<link>
        ether 8e:2e:4f:d9:04:5f  txqueuelen 1000  (Ethernet)
        RX packets 45  bytes 10639 (10.3 KiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 79  bytes 6345 (6.1 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10<host>
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

The dhcp just worked well.I havent tested udp connections yet,and I also dont know much about how to test it. p.s. : now the VMA worked well on sdf.Great!

I just tried curl --output ip.sb.out ip.sb --http0.9 -m 20,but the output is:

 /root__curl --output ip.sb.out ip.sb --http0.9 -m 20      05/09/2023 09:47:39 AM
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   602    0   602    0     0     30      0 --:--:--  0:00:20 --:--:--     0
curl: (28) Operation timed out after 20018 milliseconds with 602 bytes received
/root__ls                                                 05/09/2023 09:48:07 AM
_───_───────────_──────_───────_──────────_
│ # │   name    │ type │ size  │ modified │
├───┼───────────┼──────┼───────┼──────────┤
│ 0 │ ip.sb.out │ file │ 602 B │ now      │
_───┴───────────┴──────┴───────┴──────────_
/root__cat ./ip.sb.out                                    05/09/2023 09:48:09 AM
TTP/1.1 200 OK
Date: Tue, 09 May 2023 01:47:50 GMT
Content-Type: text/plain
Content-Length: 13
Connection: keep-alive
Cache-Control: no-cache
CF-Cache-Status: DYNAMIC
Report-To: {"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v3?s=%2B9y26TUxB6H%2FDgXjYX%2Fqi1JhdVY9i3aeAMhRmG6SPz8mgvaDPSuw8pk0cJ3tpiAhzive8aTSb3E7kP1BNUmkLCEFn9ZwlvFTcqNG%2BKEi511bq5ceN2Y8"}],"group":"cf-nel","max_age":604800}NEL: {"success_fraction":0,"report_to":"cf-nel","max_age":604800}
Server: cloudflare
CF-RAY: 7c4646d5be7608a1-SEA
alt-svc: h3=":443"; ma=86400, h3-29=":443"; ma=86400

205.166.94.4

It seems that curl can`t detect the end of the connection.

LekKit · 2023-05-09T04:42:57Z

The 'H' of "HTTP/1.1" disappeared

Ideally I'd need a full tcpdump during this curl invocation, thanks. Will look what it might be in the meantime.

WARN: net_poll_add() failed!

Oh, this is unexpected. Usually this indicates host resource exhaustion and only seen to happen on Windows with >1024 open connections. It is unexpected from kqueue to act like this, and this may also be the reason of connections not being closed.
Can you try commenting out this line and re-testing TCP?

RVVM/src/networking.c

Line 60 in 226330d

#define KQUEUE_NET_IMPL

LekKit · 2023-05-09T08:44:23Z

Weird... kqueue udata is not a void* in NetBSD???

fish4terrisa-MSDSM · 2023-05-09T09:12:01Z

The 'H' of "HTTP/1.1" disappeared

Ideally I'd need a full tcpdump during this curl invocation, thanks. Will look what it might be in the meantime.

WARN: net_poll_add() failed!

Oh, this is unexpected. Usually this indicates host resource exhaustion and only seen to happen on Windows with >1024 open connections. It is unexpected from kqueue to act like this, and this may also be the reason of connections not being closed. Can you try commenting out this line and re-testing TCP?

RVVM/src/networking.c

Line 60 in 226330d

#define KQUEUE_NET_IMPL

Commented . It worked!!! Both bug is fixed.Great.Maybe a port of NetBSD can be added.

fish4terrisa-MSDSM · 2023-05-09T09:13:29Z

Yes...I got that warning when compling RVVM on sdf.

LekKit · 2023-05-09T09:32:11Z

Yes...I got that warning when compling RVVM on sdf.

This basically means NetBSD violates kqueue API, and either needs workarounds or will have to use slower select implementation which is already used on Windows and Haiku as a safe generic implementation. Commenting out KQUEUE_NET_IMPL disables kqueue.

LekKit · 2023-05-09T11:26:06Z

Fixed kqueue issues on NetBSD in 2aba896. Report if something isn't right still.

LekKit · 2023-05-10T12:31:01Z

Fixed another (rare) kqueue bug, perhaps it hasn't bothered anyone yet but there it is

LekKit · 2023-05-12T09:46:46Z

Weird warning in lynx browser. May be related to FIN/RST flags handling.

fish4terrisa-MSDSM · 2023-06-17T03:28:41Z

Now RVVM seems not support multi task downloading (I dont know the right way to describe that),will RVVM support it?

(As it showed , the other download task is rather slow except one)

LekKit · 2023-06-17T06:41:17Z

Now RVVM seems not support multi task downloading

I am not sure at all that it's on RVVM side. I occasionally see it's balancing multiple downloads properly, sometimes not, and I have no idea what could be going wrong (I am just pumping data from a socket into ethernet frames).
It might be just that the download speed is peaking on one download, and other ones actually are received at that speed on the host.

However, running iperf in clean conditions over local network seems to distribute bandwidth evenly:

I will examine if, for example it accidentally prioritizes one socket over others.
You could also try to comment out this and re-test:

RVVM/src/networking.c

Line 54 in 5ab8066

#define EPOLL_NET_IMPL

LekKit · 2023-06-18T09:43:27Z

Running with epoll on Illumos. Apparently it balances the downloads fine, and it still peaks at around 4.5 MB/s downlink - which makes me believe this is actually peak of the Arch RISC-V mirrors capabilities (per IP?).
So far I am almost convinced this is related environment (mirrors/host) rather than RVVM.

fish4terrisa-MSDSM · 2024-01-07T09:32:25Z

TCP port forwarding works. However, there is an issue when connecting directly to 127.0.0.1 from the host, the guest gets a connection from 127.0.0.1 as well and gets confused, so the connection is dropped. I don't know how to solve that correctly yet. Will be upstreamed soon.

Is this problem fixed now? And, there isn't an option matched with port forward, maybe an option like rvvm -hostfwd ${GUEST_PORT}:${HOST_PORT} can be added? That may enable me to mount the guest fs to the host(or a 9pfs like qemu might be a better choice, but currently RVVM didn't support it) And, although nearly nobody need udp port forwarding, -tcpfwd and -udpfwd might be a better choice?

LekKit · 2024-01-19T09:49:13Z

I've almost finished a fix for localhost connections via port forwarding. Also working on distinguishing graceful FIN versus RST from outside. Will merge soon

LekKit · 2024-02-17T21:19:49Z

Please test 89a01de, it should implement all the desired (so far) functionality (But no special arguments yet).
You can ssh into the guest with ssh root@localhost -p 2022.

LekKit · 2024-02-27T22:45:48Z

Fixed a bunch of minor networking issues in staging branch.

LekKit · 2024-02-29T08:34:18Z

@X547 In upcoming patch it will be possible to have a separate IP address for the guest, so basically a non-NAT scenario without tap_linux being used.

It relies on

Being able to create multiple IP addresses on one adapter
Being able to bind tap_user sockets (both UDP/TCP) on that IP (See networking.h API for details)

The guest will be accessing the network from that address now on. It doesn't need to be a physical adapter either, so you can seat the guest onto a veth/tap/whatever interface you want. Firewall rules also can work specifically on this guest.

It is then possible to disable emulated NAT in tap_user and directly map it's IP to the assigned one, and map any ports directly without port translation (Which should be configured separately)

IPv6 is also worked on, the remaining bits are ICMPv6 and proper guest addr assignment. I am not sure yet if above technique will work for IPv6 ranges, but at least I'm sure /128 addresses will work.

3: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether e4:a7:a0:e7:25:2a brd ff:ff:ff:ff:ff:ff
    inet 192.168.0.203/24 brd 192.168.0.255 scope global dynamic noprefixroute wlan0
       valid_lft 40563sec preferred_lft 40563sec
    inet 192.168.0.37/24 scope global secondary wlan0
       valid_lft forever preferred_lft forever

LekKit · 2024-02-29T08:43:54Z

When running in unprivileged environment without additional configuration, IPv6 will be still usable for internet access, but will be NATed - there is pretty much no other option in unprivileged case

fish4terrisa-MSDSM · 2024-03-01T04:02:54Z

A separate IP address would be great, but is it possible to choose the ip address to bind with? And, it may cause ip address conflict if the user is not sure about whether the ip address is available in the network.
I think port forwarding is still useful, and options like -udpfwd and -tcpfwd would be great to see. Or, maybe a new option like -ip will help, and we may test whether the ip address is available before start the machine.

LekKit · 2024-03-01T07:28:26Z

A separate IP address would be great, but is it possible to choose the ip address to bind with? And, it may cause ip address conflict if the user is not sure about whether the ip address is available in the network. I think port forwarding is still useful, and options like -udpfwd and -tcpfwd would be great to see. Or, maybe a new option like -ip will help, and we may test whether the ip address is available before start the machine.

Basically there are going to be 2 options:

Default: Guest simply has access to internet (can download stuff, etc) and is behind a virtual NAT (So it's not recognized as a separate network host from the outside point of view). You can use port forwarding to map some host ports to guest ports. Works on all OSes without special privileges.
Interface passthrough: Guest sits on a separate address/interface so it is actually treated like a different machine in your LAN or even outside world in case of IPv6. Needs additional privileges, and unsure if it can be done on Windows. This is still WIP.

The latter option will be useful for more advanced usecases like firewalling the guest using host tools (iptables/netfilter/ipfw), accessing it as separate host without NAT, creating a virtual lan or putting it behind a VPN/proxy. It doesn't break simpler usecases (Which is to just get working internet everywhere)

fish4terrisa-MSDSM · 2024-03-01T09:40:21Z

Basically there are going to be 2 options:

Default: Guest simply has access to internet (can download stuff, etc) and is behind a virtual NAT (So it's not recognized as a separate network host from the outside point of view). You can use port forwarding to map some host ports to guest ports. Works on all OSes without special privileges.

Great. That will make my current use unaffected. Port forwarding will really help.

Interface passthrough: Guest sits on a separate address/interface so it is actually treated like a different machine in your LAN or even outside world in case of IPv6. Needs additional privileges, and unsure if it can be done on Windows. This is still WIP.

That seems attractive for me. Can you tell me which privileges (or, capabilities) in linux are needed?

The latter option will be useful for more advanced usecases like firewalling the guest using host tools (iptables/netfilter/ipfw), accessing it as separate host without NAT, creating a virtual lan or putting it behind a VPN/proxy. It doesn't break simpler usecases (Which is to just get working internet everywhere)

Hmm... It will also make it possible to use tcpdump to monitor the guest.

LekKit · 2024-03-01T09:47:55Z

That seems attractive for me. Can you tell me which privileges (or, capabilities) in linux are needed?

CAP_NET_ADMIN for adding an address to interface; CAP_NET_BIND_SERVICE if you want to make guest ports below 1024 accessible without mapping them to a different port externally.

Systems without epoll or kqueue support won't be able to bind port ranges efficiently.

LekKit · 2024-03-03T12:51:47Z

Implemented -portfwd CLI argument in c92727e.

Supported syntax:
-portfwd tcp/2022=22 - Forward TCP 0.0.0.0:2022 to guest port 22 (DHCP guest address is assumed)
-portfwd tcp/127.0.0.1:2022=22 - Forward TCP 127.0.0.1:2022 to guest port 22 (DHCP guest address is assumed)
-portfwd 25565=25565 - Forward both UDP & TCP 0.0.0.0:25565 to the same guest port
-portfwd udp/[::1]:6121=[fd9c:350d:af07::385]:6121 - Forward UDP [::1]:6121 to specific IPv6 address in guest network

LekKit added the enhancement New feature or request label Apr 10, 2023

fish4terrisa-MSDSM changed the title ~~Asking for guest port binding~~ tap_user usability tracking Apr 10, 2023

LekKit added the discussion Debate for improvement label Mar 25, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

tap_user usability tracking #87

tap_user usability tracking #87

fish4terrisa-MSDSM commented Apr 10, 2023

LekKit commented Apr 10, 2023 •

edited

Loading

LekKit commented Apr 10, 2023 •

edited

Loading

LekKit commented Apr 10, 2023 •

edited

Loading

fish4terrisa-MSDSM commented Apr 10, 2023

fish4terrisa-MSDSM commented Apr 11, 2023

fish4terrisa-MSDSM commented Apr 11, 2023

LekKit commented May 6, 2023

LekKit commented May 7, 2023 •

edited

Loading

LekKit commented May 8, 2023

fish4terrisa-MSDSM commented May 9, 2023

fish4terrisa-MSDSM commented May 9, 2023

LekKit commented May 9, 2023 •

edited

Loading

LekKit commented May 9, 2023

fish4terrisa-MSDSM commented May 9, 2023

fish4terrisa-MSDSM commented May 9, 2023

LekKit commented May 9, 2023 •

edited

Loading

LekKit commented May 9, 2023

LekKit commented May 10, 2023

LekKit commented May 12, 2023

fish4terrisa-MSDSM commented Jun 17, 2023

LekKit commented Jun 17, 2023

LekKit commented Jun 18, 2023

fish4terrisa-MSDSM commented Jan 7, 2024 •

edited

Loading

LekKit commented Jan 19, 2024 •

edited

Loading

LekKit commented Feb 17, 2024

LekKit commented Feb 27, 2024

LekKit commented Feb 29, 2024

LekKit commented Feb 29, 2024 •

edited

Loading

fish4terrisa-MSDSM commented Mar 1, 2024

LekKit commented Mar 1, 2024 •

edited

Loading

fish4terrisa-MSDSM commented Mar 1, 2024

LekKit commented Mar 1, 2024 •

edited

Loading

LekKit commented Mar 3, 2024 •

edited

Loading

tap_user usability tracking #87

tap_user usability tracking #87

Comments

fish4terrisa-MSDSM commented Apr 10, 2023

LekKit commented Apr 10, 2023 • edited Loading

LekKit commented Apr 10, 2023 • edited Loading

Current tap_user issues/TODOs:

LekKit commented Apr 10, 2023 • edited Loading

fish4terrisa-MSDSM commented Apr 10, 2023

fish4terrisa-MSDSM commented Apr 11, 2023

fish4terrisa-MSDSM commented Apr 11, 2023

LekKit commented May 6, 2023

LekKit commented May 7, 2023 • edited Loading

LekKit commented May 8, 2023

fish4terrisa-MSDSM commented May 9, 2023

fish4terrisa-MSDSM commented May 9, 2023

LekKit commented May 9, 2023 • edited Loading

LekKit commented May 9, 2023

fish4terrisa-MSDSM commented May 9, 2023

fish4terrisa-MSDSM commented May 9, 2023

LekKit commented May 9, 2023 • edited Loading

LekKit commented May 9, 2023

LekKit commented May 10, 2023

LekKit commented May 12, 2023

fish4terrisa-MSDSM commented Jun 17, 2023

LekKit commented Jun 17, 2023

LekKit commented Jun 18, 2023

fish4terrisa-MSDSM commented Jan 7, 2024 • edited Loading

LekKit commented Jan 19, 2024 • edited Loading

LekKit commented Feb 17, 2024

LekKit commented Feb 27, 2024

LekKit commented Feb 29, 2024

LekKit commented Feb 29, 2024 • edited Loading

fish4terrisa-MSDSM commented Mar 1, 2024

LekKit commented Mar 1, 2024 • edited Loading

fish4terrisa-MSDSM commented Mar 1, 2024

LekKit commented Mar 1, 2024 • edited Loading

LekKit commented Mar 3, 2024 • edited Loading

LekKit commented Apr 10, 2023 •

edited

Loading

LekKit commented Apr 10, 2023 •

edited

Loading

Current `tap_user` issues/TODOs:

LekKit commented Apr 10, 2023 •

edited

Loading

LekKit commented May 7, 2023 •

edited

Loading

LekKit commented May 9, 2023 •

edited

Loading

LekKit commented May 9, 2023 •

edited

Loading

fish4terrisa-MSDSM commented Jan 7, 2024 •

edited

Loading

LekKit commented Jan 19, 2024 •

edited

Loading

LekKit commented Feb 29, 2024 •

edited

Loading

LekKit commented Mar 1, 2024 •

edited

Loading

LekKit commented Mar 1, 2024 •

edited

Loading

LekKit commented Mar 3, 2024 •

edited

Loading