These are chat archives for SmingHub/Sming

17th
Dec 2015
robotiko
@robotiko
Dec 17 2015 12:38
@hreintke .. Dev is now Sming 2.0?
any pending PR to be included? there is good number of them
alon24
@alon24
Dec 17 2015 12:40
so I am now reinstalling my home computer with win10, what is the correct way to add sming, now to pure win10?
Curtis Pope
@piperpilot
Dec 17 2015 12:41
@hreintke I just tested PR 489 and it seems to fix the issue I was seeing...I'll do further testing in a little while after I get the kids to school, but so far it seems to be good! We need to make sure its included in 2.0
hreintke
@hreintke
Dec 17 2015 14:02
@robotiko :
I know there are a number of PR's still outstanding.
However, if we include them now, we again are delaying the release of V2.0 as there should be some time/testing between including in develop and release to master. I prefer to release now in include the mentioned PR's to develop directly afterwards.
@piperpilot :
The above is also true for PR 489. It is a conceptual change which even needs additional updates to keep current functionality. Would like to see it in use some time in develop to know there are no unexpected side effects.
robotiko
@robotiko
Dec 17 2015 14:07
@hreintke that includes HW timer and servo lib?
Curtis Pope
@piperpilot
Dec 17 2015 14:27
@hreintke while I agree, we are releasing 2.0 with a known serious bug...that I can't agree with.
hreintke
@hreintke
Dec 17 2015 14:28
@robotiko : Yes, HW Timer & Servo lib have been available as PR but not yet merged in Develop.
I was busy in another area which prevented me to be really active on merging PR's but also was trying to get to a "quiet" situation for releasing.
hreintke
@hreintke
Dec 17 2015 14:35
@piperpilot : I know it is a serious bug but
  • Until a couple of days ago, no-one even knew it existed -> the actual impact is limited.
  • Only you, and then even limited, has tested the PR.
  • The update is a very conceptual one which might effect other parts.
  • Also knowing the PR's @robotiko is mentioning, the time to release V2.1 will be not be to far away.
Curtis Pope
@piperpilot
Dec 17 2015 14:39
It is bad practice to release with a known regression. There are probably other users out there that use hardware serial. It works fine on master but is broken on the new release. It was only discovered recently because it is a new regression within the last 2 weeks and there are only a few of us testing develop.
It doesn't impact me because I can use develop or even my own patched version but it will impact other users
Alexander V. Ribchansky
@avr39-ripe
Dec 17 2015 15:06
@hreintke @piperpilot Hello! Do we have some strict time to release new version? Can we just hold a while, and test mentioned PR, and then either release without it with some errata in release notes or merge it and whole world will be happy ;o) ?
Curtis Pope
@piperpilot
Dec 17 2015 15:11
@avr39-ripe that was kind of my suggestion...since we have a known regression, we should hold the release
Alexander V. Ribchansky
@avr39-ripe
Dec 17 2015 15:14
@piperpilot I agree.. If I do not miss something important, in meanwhile we make Sming just for "ourselves", we are not kind of corporation with deadlines and so on.. so just holding release doesn't kill anybody. One who is advanced Sming-er can checkout develop and/or adopt PR and so on.. regular users have nice working master.. Everybody happy ;)
hreintke
@hreintke
Dec 17 2015 15:59

@piperpilot @avr39-ripe :
At least two major issues are solved in the current Develop release :

  • Memory leakage which effects all users of the arduinoJson
  • Since the release of UDK 209, all new windows users of Sming, which follow the installation procedure from Github will have a non working environment.

When we "hold a while" (which we did f.e on december 3rd when UDK 209 was release and the decision was made to postpone release to include SDK 1.5.0 version) there will always be a reason for not releasing.
Regarding the PR : It is not finished PR, just a rough implementation and, as I just see in the PR, there are already side effects of the PR -> needs investigation and updating.
I rather have a released version with a "known bug" than a version with possible unknown issues ( I know who will need to investigate :smile: )
The "Solution in the middle" is : Release V2.0 now, only merge Serial fix into develop, evaluate and test, when OK release V2.1.

Curtis Pope
@piperpilot
Dec 17 2015 16:52
Just realize if you release, it is broken serial and no way to roll back/fix for users. Like I said, no problem for me as I can use develop...I have had to for a while due to the Json issue, but it might break others. If we can get 2.1 out quickly, then fine, but can't wait for weeks with a known critical bug.
robotiko
@robotiko
Dec 17 2015 17:18
@hreintke @piperpilot @avr39-ripe
I do agree with @hreintke that it is better to have a consistent release kwnoing that contains a kwnon issue.. that introduce a half baked fix
that might force to roll back version
I don't knwo If i have been hit by that issue too .. becuase sometimes it simply gets weird and you get used to it :D
what I don't have clear .. is the PR merge policy
some PR have been tested .. modified code according the sming team request.. retested.. and are waiting.. while some other fixed.. are merged rigth way
robotiko
@robotiko
Dec 17 2015 17:43
to esp32 boys.. in case you didn't see http://esp32.de/
has a lot of info.. all communications
unboxing.. video.. pinouts
pcb.. etc
the sdk 1.0 doc is out
Curtis Pope
@piperpilot
Dec 17 2015 17:45
very cool
wonder how long before it comes out mainstream
hreintke
@hreintke
Dec 17 2015 17:53
nice, rtos doc is very similar to esp8266 rtos doc
robotiko
@robotiko
Dec 17 2015 17:54
got it from repo?
forum .. i mean
hreintke
@hreintke
Dec 17 2015 17:54
yes
robotiko
@robotiko
Dec 17 2015 17:54
ok
so you have been chewing that doc recently
hwo similar are both devices rtos apis?
hreintke
@hreintke
Dec 17 2015 17:56
no not chewing, just first scan and done in 5 minutes
robotiko
@robotiko
Dec 17 2015 18:01
:D
zhivko
@zhivko
Dec 17 2015 18:06
Thanks for esp32 links! Hope we could support this device also... Somehow..
hreintke
@hreintke
Dec 17 2015 18:14
@robotiko :
agree on better PR merge policy. I do see the same issues like you.
Main question for me is : how is clear what the status of a PR is.
Can be improved by using labels/milestones.
@all : If there are ideas proposals -> share
Curtis Pope
@piperpilot
Dec 17 2015 19:35
what is the latest version of esptool.py that works with windows...do we have our own repo for it? I have another developer I work with that I want to flash some code and try it out...but he is on windows
are we using the standard esptool.py from here: https://github.com/themadinventor/esptool
robotiko
@robotiko
Dec 17 2015 20:54
@hreintke Thsi kidn of projects usually tag with a milestone and the Issue type (bug, improvement, new functionality) (github doesn't offer much more ) we could also think about adding priority (low, normal, high ,critical)
right now .. there is not much people testing..
but if something is already tested.. the inclusion or not in current release should be based on "stability or release delay " parameters
having a non Merged PR for long time.. many times includes bgfixing on PR to be consistent with laetr developments .. and for sure.. testing again.
so the sooner.. the better to merge tested stuff
when gdbstub fixed, we should try
it has xtensa-lx106-gdb.exe
robotiko
@robotiko
Dec 17 2015 21:06
coooool!
thanks a lot
master!
alonewolfx2
@alonewolfx2
Dec 17 2015 21:07
@robotiko did you test it?
i didnt make it work
Dmitry Kireev
@kireevco
Dec 17 2015 21:07
as well is esptool2
so, I have a theory about incompatability
robotiko
@robotiko
Dec 17 2015 21:08
@alonewolfx2 no way to test it so far .. the gdb PR doesnt compile
Dmitry Kireev
@kireevco
Dec 17 2015 21:08
@alonewolfx2 we need to remove include <c_types.h> from gdbsub or #include <stdint.h>
please try this dirty fix and see how it goes. @alonewolfx2 I know you like dirty hacking :)
something is wrong with gdb.exe
Dmitry Kireev
@kireevco
Dec 17 2015 21:13
I need someone to test it please