-->
Page 2 of 4

Re: LWIP to V2

PostPosted: Tue Sep 21, 2021 12:37 pm
by eriksl
That would be absolutely my dream. Espressif wants us all to first move to RTOS and when there, "seemlessly" move to ESP32. But the ESP32 has way too much features for me, but more important, their SDK shields off all hardware access or possibilities for own OTA implementations. And then the form factor, you either must use a "development" board which is way too large for me to be useful (I am now using ESP12's) or a "production" board which has the contacts very close to each other, very hard to solder right and probably impossible to use with pin arrays like I now do (so you replace it without soldering).

So all I want Espressif to do is to just disclose ALL information, information they have, about the hardware and about the SDK (for example about OTA). I understand they cannot disclose the WLAN part, but that's relatively small. Most of the now non-disclosed code has nothing to do with RF.

And once all information is available, I'd happily make my own RTOS-like environment, but then completely open. I don't understand why they're so unwilling.

Re: LWIP to V2

PostPosted: Tue Sep 21, 2021 12:44 pm
by eriksl
BTW current state of of work is in https://github.com/eriksl/lwip-for-esp8266-nonos-sdk, branch master. This branch will be rebased very frequently, so to have the latest version, reset your branch to somewhere way back in history and then pull, instead of just pull.

I've found mysterious issues that simply made all IP stop working and then I had to "rebase back" a lot (and that's not simple work) and then dissect the commit that causes the issue. Currently I am at one. To prevent having to "rebase back" (which can cause my own commits to become damaged), I now make a branch of every successful rebase. Currently that is 00-72dcdb76558c4c123f4973acf8d9918457a25c09 and it's currently equal to master. If last rebase doesn't work, I can compare with the latest working status and rollback by just throwing away master and copy that branch to master. This also means the last commit I succesfully rebased was indeed 72dcdb76558c4c123f4973acf8d9918457a25c09. Still a long way to go.

It seems the commit following this one, which breaks lwip for us, introduces a concept where netif's don't need to be set "up" by DHCP so they don't. But the SDK code relies on this. So I will try and see what happens if I revert that specific code. Whether both SDK code and LWIP are still happy. Otherwise it will end there :?

Re: LWIP to V2

PostPosted: Tue Sep 21, 2021 8:10 pm
by davydnorris
eriksl wrote:That would be absolutely my dream. Espressif wants us all to first move to RTOS and when there, "seemlessly" move to ESP32. But the ESP32 has way too much features for me, but more important, their SDK shields off all hardware access or possibilities for own OTA implementations. And then the form factor, you either must use a "development" board which is way too large for me to be useful (I am now using ESP12's) or a "production" board which has the contacts very close to each other, very hard to solder right and probably impossible to use with pin arrays like I now do (so you replace it without soldering).

So all I want Espressif to do is to just disclose ALL information, information they have, about the hardware and about the SDK (for example about OTA). I understand they cannot disclose the WLAN part, but that's relatively small. Most of the now non-disclosed code has nothing to do with RF.

And once all information is available, I'd happily make my own RTOS-like environment, but then completely open. I don't understand why they're so unwilling.


Absolutely in sync with you there - I use the official Espressif WROOM modules now, but have used the AIThinker ESP-12 modules previously. I need the ones with most certification - makes it easier to get Class Compliance certification on my own devices.

I have a bunch of 32's but the 8266 does what I need and is lower power consumption and cheaper.

Re: LWIP to V2

PostPosted: Wed Sep 22, 2021 10:10 am
by eriksl
I think 90% of the ESP8266 users are not even using 10% of the capabilities of it. And that's how Espressif wants to have it, so they can sell more ESP32's :?

I can imagine you need to have certifications, for me that's not important. For me it's the size and the ESP12F's and ESP12S's are just perfect for that. It would be even better if someone would come up with an even smaller print size and not two but three or four pin rows to accomodate, no problem for me.

Even though I wouldn't object if the ESP32 would be "opened up" completely instead. I just need to find a solution for the very small pin-land-area pitch. I'd love to have I2C and PWM in hardware, even though the processor is capable enough to do it all in software, no problem.