These are chat archives for esp8266/Arduino

2nd
May 2017
Ivan Grokhotkov
@igrr
May 02 2017 06:59
@Makuna regarding "Why is the github repository set so that I have to fork/branch the project into my repo before creating a pull request?" — this isn't a setting that I am aware of... if you instruct me what i need to change in the repository settings, i will.
In general, i don't mind adding you as a collaborator to the repository if that helps.
Michael Miller
@Makuna
May 02 2017 07:02
Does it require that you have to make people a contributor to just clone and then submit a pull? I don't know if you spotted it, but I can submit a pull directly from my local to the origin and not have to go through my fork on github.
So this makes it easy enough.
Ivan Grokhotkov
@igrr
May 02 2017 07:03
AFAIK you still need to have the PR branch somewhere on github in order to submit a PR.
This branch can either live in the project repo itself (esp8266/Arduino), or in some fork of it.
If you are a collaborator, you can push new branches to the project repo, without having to create a fork.
Otherwise, you need to have a fork in order to store the PR branch somewhere.
Michael Miller
@Makuna
May 02 2017 07:05
So what I did, Fork to my Github, Clone to my desktop, make changes locally. Create a Pull from local directly to esp8266/Arduino without putting it into my github fork.
Ivan Grokhotkov
@igrr
May 02 2017 07:05

submit a pull directly from my local to the origin

I haven't noticed such a function, are you using Github Desktop client?

Michael Miller
@Makuna
May 02 2017 07:06
yes
Ivan Grokhotkov
@igrr
May 02 2017 07:06
maybe that is some magic it does then. i don't know how to do that in a generic git command line way...
Michael Miller
@Makuna
May 02 2017 07:06
If you look at the one I created for the PROGMEM issue, you will see it.
Me No Dev
@me-no-dev
May 02 2017 07:07
@Makuna @igrr commit protection is on on the master branch
maybe that is the reason
Michael Miller
@Makuna
May 02 2017 07:07
My Git expert said you should be able to go from any one location to any other location as long as they had some relationship.
Merge protection should be on I would expect, is this the same as commit?
Me No Dev
@me-no-dev
May 02 2017 07:09
you can not work directly on the master. Have to make a branch and merge that instead
at least that is how I understand it
Michael Miller
@Makuna
May 02 2017 07:09
Its not a big deal (anymore), it just adds "cliffs" for people who might be interested in participating.
Branch yes, but why a fork? Today I have to make a fork.
Ivan Grokhotkov
@igrr
May 02 2017 07:10
Yes, but we haven't been touching on branch protection related issues... just the question of "can i make a PR without making a fork"?
What i think Github Desktop client does is that it actually forks and creates a temporary branch for you, then pushes to that branch.
Michael Miller
@Makuna
May 02 2017 07:13
Your second link sums up my complaint. I have to have a bunch of repros forked just to make pushes and it makes it cluttered and seem like I am supporting them from my account.
Me No Dev
@me-no-dev
May 02 2017 07:13
wait... why is that a probem? I thought that is how it should work. You should need your own copy of it all in order to contribute and unless you are a member of the repository, you do need a fork
I do get your furstration though
my account looks the same
Michael Miller
@Makuna
May 02 2017 07:14
I just want people to not find my forks (because my other work is valid and they think the forks are valid).
Me No Dev
@me-no-dev
May 02 2017 07:14
I agree, but I doubt there is a way around that\
Michael Miller
@Makuna
May 02 2017 07:14
I did have someone fork my fork and submit a pull to my account!
yeah, it seems like something we have to live with for now. But at least I can remove the one step of submitting the pull to my Github fork.
Ivan Grokhotkov
@igrr
May 02 2017 07:22
Folks, any thoughts on setting up a help/discussion forum for ESP8266 Arduino?
Originally my idea was that esp8266.com would be used as a help/troubleshooting platform, but nowadays the forum is barely usable (especially on mobile), so i was wondering if it might make sense to set up something user-friendly, based on Discourse or similar.
Bert Melis
@bertmelis
May 02 2017 08:00
I'm in favour for it. But then I'll have a GH-account, a gitter account, an esp8266-forum account, and esp32.com account and the "new esp8266 Arduino forum"-account. And they all are related. Consolidation to some extent would be welcome
Ivan Grokhotkov
@igrr
May 02 2017 08:01
That's a very good point.
We have been discussing consolidating bbs.espressif.com and esp32.com, but both of these are Espressif's support/E2E forums. Since this project is not affiliated with Espressif (except for the fact that two developers work there), and i would prefer to keep the discussion forum/board separate as well.
Bert Melis
@bertmelis
May 02 2017 08:04
don't get me wrong: github is not the place for help on using the framework. gitter can, but has the limitations of a chat room (several discussions at once). So a seperate discussion board is imho the way to go
Ivan Grokhotkov
@igrr
May 02 2017 08:04
Discourse does support "log in via Github", so that should take at least some pain away
Bert Melis
@bertmelis
May 02 2017 08:05
that's a plus! It avoids the confusion of different nicknames
Ivan Grokhotkov
@igrr
May 02 2017 08:05
Yes, gitter and chats in general are very attention-hungry. They don't let one to respond by email at some convenient time. So I would prefer to have something less interactive for a support forum.
(following that line, i should probably move this discussion from gitter into a github issue)
Ivan Grokhotkov
@igrr
May 02 2017 12:47
Request for comment: igrr/axtls-8266#44 (support for minimal root certificate store)
rodrigocirilo
@rodrigocirilo
May 02 2017 13:55
Hello friends.
Could you help me on another question.
How to save in eeprom of arduino nano or nodemcu, a TAG of RFID card? And the number that retreats is generally thus an example: 7A 12 E5 90 being a string.
And how many eeprom addresses does the nodemcu have?
riataman
@riataman
May 02 2017 13:57
igrr: Hey, how do you envision the API for this one: esp8266/Arduino#1062 ?
I have code sitting on my machine to make it work. I'm thinking of just introducing a SPI.swap() call similar to Serial.swap()
Ivan Grokhotkov
@igrr
May 02 2017 14:16
That's pretty cool! Given that we can't use two SPIs at the same time (since there is only one SPI controller) that does make sense.
riataman
@riataman
May 02 2017 14:16
Stil is of limited use.
I wanted to use it with a SD card, but the card gets confused by all the taking over from the flash.
Ivan Grokhotkov
@igrr
May 02 2017 14:18
However there's one thing I don't like about swap (for UART as well) is that the new state depends on the old one... A method like .pins(miso, mosi, sck, ss) would be a better API.
riataman
@riataman
May 02 2017 14:18
SPI.setOverlapMode() ?
Ivan Grokhotkov
@igrr
May 02 2017 14:18
And if we eventually have a bitbang SPI then we can choose the impl to use depending on the pins selected.
Why is SD getting confused though, doesn't it have a dedicated CS line?
riataman
@riataman
May 02 2017 14:19
well
first you have to use hardware CS, you can't use a GPIO for CS
because you don't actually know when the flash is going to take over
and then the SD card doesn't seem to play nicely in a shared SPI bus
I checked this all with a logic analizer, the spi commands where the same in the normal HSPI pins vs the overlap mode
Ivan Grokhotkov
@igrr
May 02 2017 14:21
That's the part I don't understand. In SPI mode SD card should be okay sharing the bus with other devices, as long as it has its own CS. At least that has been my past experience...
Maybe need to put an SD card emulator in there and see.
riataman
@riataman
May 02 2017 14:22
does those exist?
there are a few things that are going on
first CS is GPI0
only hardware CS available
and that pulses during boot for some reason
the SD card certainly didn't like that
so I resorted to inserting the SD card AFTER boot
I got it to respond to me after that, went into IDLE mode and responded the version
but at the first time the flash interrupted the SPI stream
it broke again
even when the CS line was correctly high
Ivan Grokhotkov
@igrr
May 02 2017 14:26
Ah, so the flash read kicks in in the middle of the transfer?
riataman
@riataman
May 02 2017 14:26
yeah
I changed a lot of code
to use block transfer instead of byte by byte
Ivan Grokhotkov
@igrr
May 02 2017 14:27
Yes, these exist, have one running on an FPGA.
riataman
@riataman
May 02 2017 14:27
which the SD library does
I can setup this again and send you screenshoots of logic analizer with what I'm seeing.
Ivan Grokhotkov
@igrr
May 02 2017 14:28
Okay. This can possibly be worked around by placing transfer function into IRAM, so that it doesn't trigger cache refills
riataman
@riataman
May 02 2017 14:28
I tried that too ...
but probably didn't went far enough with it
Ivan Grokhotkov
@igrr
May 02 2017 14:28
At any rate, I think it may be worth putting this feature in, even there are some issues with SD card.
riataman
@riataman
May 02 2017 14:29
yes
Ivan Grokhotkov
@igrr
May 02 2017 14:29
Can you make a PR with .pins function added?
riataman
@riataman
May 02 2017 14:29
sure, I actually want to test it with some other device before summiting
Ivan Grokhotkov
@igrr
May 02 2017 14:30
Fair enough, you can also possibly submit it and someone else can test. I am sure many folks have SPI LCDs at their disposal.
riataman
@riataman
May 02 2017 14:30
I was testing this with a 23LC1024 yesterday
but couldn't make it work
couldn't even make it work with the normal HSPI pins so I was doing something wrong
I'll try to massage this into shape, it is very little code.
I need all of this because I wanted to make a i2s recorder
also got i2s input to work
something else to be nice to summit
I got it to work using the SD card and SoftwareSPI, but its not fast enought for 44.5Khz 2channel 16bit sound
I can record about 50Mb but it keeps falling behind in the DMA buffers (40kb worth of buffers)
that's why I wanted to get the hspi overlap mode to work. Also because software SPI actually leaves me with no pins left.
riataman
@riataman
May 02 2017 14:36
(i2s input uses the same HSPI pins that's why I can use it in normal mode).
riataman
@riataman
May 02 2017 19:00
ok, I finally got that 23LC1024 to work in the alternative pins
I convert it all to use single SPI writes/transfers instead of a few individual ones.
may need to do the same with the SD card
riataman
@riataman
May 02 2017 20:44
Ok, I pushed that: esp8266/Arduino#3189 hopefully I didn't screw up that badly on the git side of things.
Steve Nelson
@zenmanenergy
May 02 2017 20:54
man that's cool if that works. Does this potentially open up 4 more GPIOs when using SPI modules? Or am i reading that code wrong?
riataman
@riataman
May 02 2017 20:54
yes
Steve Nelson
@zenmanenergy
May 02 2017 20:55
well i'm crossing my fingers!
riataman
@riataman
May 02 2017 20:55
it worked for me
with a 23LC1024
an SD card didn't work
The only other spi device I have is a NRF24L01+
I haven't tried with that
probably most of the Arduino code that is out there won't work as is
as manual CS handling doesn't work. You must let the hardware handle the CS pin
All 3 of the arduino libraries that I found didn't work with the 23LC1024
because they expected to be able to activate the CS and then send several commands
and deassert it
I had to convert it to send a single command with all the bytes it needed in a single go
Steve Nelson
@zenmanenergy
May 02 2017 21:01
sounds great. Do you think it satisfies the $250 bounty? Or was that for something else?
riataman
@riataman
May 02 2017 21:02
I think that is what the $250 bounty was about.
Michal Vondráček
@majklovec
May 02 2017 23:32
Hello is it possible to use F() and PSTR() in a class? Whenever I use F() in a class I get: "cores\esp8266/pgmspace.h:21:51: note: '__c' was declared here"