net.lookup locked

[Vagner nascimento]

Hi Grabriel,

About eLua Web Build, in a couple of days the new version will be working with eLua 0.9 . We are reimplementing some modules to support the new eLua build system written in Lua.

Will be announced here soon.

Best,

Vagner

On Tue, May 14, 2013 at 9:57 PM, Gabriel Duarte <[hidden email]> wrote:

Hello Martin!

With this firmware you sent me, I cant get even access to eLua's command line, be through screen or Goat :/ This firmware is compiled with UART support?

I could not find how to build eLua 0.9 from eLua Web builder, only 0.8.

I have tested SM's 0.8, elua_lua_at32uc3a0256.hex, from mizar32-firmware-20120123.tgz and does not work anyway :(

2013/5/14 Martin Guy <[hidden email]>

> You got any idea where is the problem? uIP?

  That's a big question. It could be one of many things, and whether
it works for you is one test that may suggest the cause.
  To apply most debugging methods, you have to be able to repeat the
failure. At present I only have it failing on your network and my
colleague's home network, with you using your firmware from the Web
Builder and him using his own firmware that he has been modifying to
improve other areas.  That's why I ask you to test SM's 0.8 release
binary and eluaproject's 0.9 Mizar32 binary, reducing the problem to
bugs in those firmwares, problems with your Mizar32 hardware and the
setup of your home network.

   M
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[gduarte]

I will try to deploy this firmware tonight and I let you know what happens. I will describe which commmands I've used to either burn the firmware and access eLua's console.

One thing I can assure you is that I used screen to accees the serial with the following parameters:

screen /dev/ttyUSB0 8n1

With SM's 0.8 based firmware, it works fine, but with the 0.9 you sent me, I cant access eLua's console.

2013/5/15 Vagner nascimento <[hidden email]>

Hi Grabriel,

About eLua Web Build, in a couple of days the new version will be working with eLua 0.9 . We are reimplementing some modules to support the new eLua build system written in Lua.

Will be announced here soon.

Best,

Vagner

On Tue, May 14, 2013 at 9:57 PM, Gabriel Duarte <[hidden email]> wrote:

Hello Martin!

With this firmware you sent me, I cant get even access to eLua's command line, be through screen or Goat :/ This firmware is compiled with UART support?

I could not find how to build eLua 0.9 from eLua Web builder, only 0.8.

I have tested SM's 0.8, elua_lua_at32uc3a0256.hex, from mizar32-firmware-20120123.tgz and does not work anyway :(

2013/5/14 Martin Guy <[hidden email]>

> You got any idea where is the problem? uIP?

  That's a big question. It could be one of many things, and whether
it works for you is one test that may suggest the cause.
  To apply most debugging methods, you have to be able to repeat the
failure. At present I only have it failing on your network and my
colleague's home network, with you using your firmware from the Web
Builder and him using his own firmware that he has been modifying to
improve other areas.  That's why I ask you to test SM's 0.8 release
binary and eluaproject's 0.9 Mizar32 binary, reducing the problem to
bugs in those firmwares, problems with your Mizar32 hardware and the
setup of your home network.

   M
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[Martin Guy]

Oops, got the Cc:s wrong...

---------- Forwarded message ----------
From: Martin Guy <[hidden email]>
Date: 15 May 2013 20:30
Subject: Re: [eLua-dev] net.lookup locked
To: [hidden email]


OK, I've found one problem.

The web builder defines BUILD_DHCP but the correct define is
BUILD_DHCPC... so the firmware it produces never enables DHCP, if you
select it or not, and always uses fixed IP addresses. This explains
why it always appeared as 192.168.1.* for you!

The web builder's author has fixed this but it's still giving me
problems with DHCP and with net.lookup. From my tests I fear that we
have compiler bugs in Atmel's toolchain.

Can I ask you to try this firmware, which is working for me with DHCP
and with net.lookup() also?

Sorry about this...

    M

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[gduarte]

Cool! At least one bug found and fixed :)

As I said before, tonight I will try to test this firmware! You don't need to be sorry, it is being a pleasure to help you and improve eLua and Mizar32 ecosystem :)

2013/5/15 Martin Guy <[hidden email]>

Oops, got the Cc:s wrong...

---------- Forwarded message ----------
From: Martin Guy <[hidden email]>
Date: 15 May 2013 20:30
Subject: Re: [eLua-dev] net.lookup locked
To: [hidden email]


OK, I've found one problem.

The web builder defines BUILD_DHCP but the correct define is
BUILD_DHCPC... so the firmware it produces never enables DHCP, if you
select it or not, and always uses fixed IP addresses. This explains
why it always appeared as 192.168.1.* for you!

The web builder's author has fixed this but it's still giving me
problems with DHCP and with net.lookup. From my tests I fear that we
have compiler bugs in Atmel's toolchain.

Can I ask you to try this firmware, which is working for me with DHCP
and with net.lookup() also?

Sorry about this...

    M

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[gduarte]

Hello!

So, good news, in steps/

1 - Before, I could not access eLua's shell because I was pointing to the wrong device, ttyUSB0 instead of ttyACM0. The device changed I do know why, but fortunatelly I saw the change in dmesg's output.

2 - It works!!!!!! net.lookup could resolve names and return the correct parameters to net.connect :D

I would like to help more if you need, be programming, giving ideas or reporting bugs!

Thanks for the good job and for helping me to solve my problem, now I can continue with my studies and development!

Cheers

2013/5/15 Gabriel Duarte <[hidden email]>

Cool! At least one bug found and fixed :)

As I said before, tonight I will try to test this firmware! You don't need to be sorry, it is being a pleasure to help you and improve eLua and Mizar32 ecosystem :)

2013/5/15 Martin Guy <[hidden email]>

Oops, got the Cc:s wrong...

---------- Forwarded message ----------
From: Martin Guy <[hidden email]>
Date: 15 May 2013 20:30
Subject: Re: [eLua-dev] net.lookup locked
To: [hidden email]


OK, I've found one problem.

The web builder defines BUILD_DHCP but the correct define is
BUILD_DHCPC... so the firmware it produces never enables DHCP, if you
select it or not, and always uses fixed IP addresses. This explains
why it always appeared as 192.168.1.* for you!

The web builder's author has fixed this but it's still giving me
problems with DHCP and with net.lookup. From my tests I fear that we
have compiler bugs in Atmel's toolchain.

Can I ask you to try this firmware, which is working for me with DHCP
and with net.lookup() also?

Sorry about this...

    M

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[gduarte]

Ahhhh just to let you know... I have built a firmware from master branch and it does not work, only the firmware you sent me.

2013/5/15 Gabriel Duarte <[hidden email]>

Hello!

So, good news, in steps/

1 - Before, I could not access eLua's shell because I was pointing to the wrong device, ttyUSB0 instead of ttyACM0. The device changed I do know why, but fortunatelly I saw the change in dmesg's output.

2 - It works!!!!!! net.lookup could resolve names and return the correct parameters to net.connect :D

I would like to help more if you need, be programming, giving ideas or reporting bugs!

Thanks for the good job and for helping me to solve my problem, now I can continue with my studies and development!

Cheers

2013/5/15 Gabriel Duarte <[hidden email]>

Cool! At least one bug found and fixed :)

As I said before, tonight I will try to test this firmware! You don't need to be sorry, it is being a pleasure to help you and improve eLua and Mizar32 ecosystem :)

2013/5/15 Martin Guy <[hidden email]>

Oops, got the Cc:s wrong...

---------- Forwarded message ----------
From: Martin Guy <[hidden email]>
Date: 15 May 2013 20:30
Subject: Re: [eLua-dev] net.lookup locked
To: [hidden email]


OK, I've found one problem.

The web builder defines BUILD_DHCP but the correct define is
BUILD_DHCPC... so the firmware it produces never enables DHCP, if you
select it or not, and always uses fixed IP addresses. This explains
why it always appeared as 192.168.1.* for you!

The web builder's author has fixed this but it's still giving me
problems with DHCP and with net.lookup. From my tests I fear that we
have compiler bugs in Atmel's toolchain.

Can I ask you to try this firmware, which is working for me with DHCP
and with net.lookup() also?

Sorry about this...

    M

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[gduarte]

The boad grabs an dynamic IP address but does not resolve with net.lookup...

What is the difference between these 2 branchs?

2013/5/15 Gabriel Duarte <[hidden email]>

Ahhhh just to let you know... I have built a firmware from master branch and it does not work, only the firmware you sent me.

2013/5/15 Gabriel Duarte <[hidden email]>

Hello!

So, good news, in steps/

1 - Before, I could not access eLua's shell because I was pointing to the wrong device, ttyUSB0 instead of ttyACM0. The device changed I do know why, but fortunatelly I saw the change in dmesg's output.

2 - It works!!!!!! net.lookup could resolve names and return the correct parameters to net.connect :D

I would like to help more if you need, be programming, giving ideas or reporting bugs!

Thanks for the good job and for helping me to solve my problem, now I can continue with my studies and development!

Cheers

2013/5/15 Gabriel Duarte <[hidden email]>

Cool! At least one bug found and fixed :)

As I said before, tonight I will try to test this firmware! You don't need to be sorry, it is being a pleasure to help you and improve eLua and Mizar32 ecosystem :)

2013/5/15 Martin Guy <[hidden email]>

Oops, got the Cc:s wrong...

---------- Forwarded message ----------
From: Martin Guy <[hidden email]>
Date: 15 May 2013 20:30
Subject: Re: [eLua-dev] net.lookup locked
To: [hidden email]


OK, I've found one problem.

The web builder defines BUILD_DHCP but the correct define is
BUILD_DHCPC... so the firmware it produces never enables DHCP, if you
select it or not, and always uses fixed IP addresses. This explains
why it always appeared as 192.168.1.* for you!

The web builder's author has fixed this but it's still giving me
problems with DHCP and with net.lookup. From my tests I fear that we
have compiler bugs in Atmel's toolchain.

Can I ask you to try this firmware, which is working for me with DHCP
and with net.lookup() also?

Sorry about this...

    M

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

--

Gabriel Duarte

Linux User #471185
Rio de Janeiro / RJ
http://genericdev.wordpress.com/

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[Martin Guy]

Hi Gabriel
  Thanks for your detailed feedback

> Before, I could not access eLua's shell because I was pointing to the wrong device, ttyUSB0 instead of ttyACM0.

OK thanks.

2 - It works!!!!!! net.lookup could resolve names and return the
correct parameters to net.connect :D

"It" == ??? -rc2? -rc1? Both? To find the bug I need to divide the
firmwares into ones that seem to work and those that are known to have
some problem.
I mean, I can test them here, but that doesn't mean that they work for you!

> I have built a firmware from master branch and it does not work, only the firmware you sent me.

OK, good to know you are seeing the same results as me. Thanks.

> What is the difference between these 2 branchs?

- The compiler (Atmel-GCC 3.4.1 on b.sm.it vs. my one built with ct-ng here)
- builder.simplemachines.it doesn't define BUILD_RTC so you get no
mizar32.rtc.*() Lua module, which was added since the web builder was
written. That "should" make no difference to net.*() but if it's the
kind of bug that corrupts memory at random, just adding one byte
changes what gets crapped on
- builder.sm.it is a 64-bit Linux, my laptop has 32-bit linux. That
should make no difference either! But I have seen that the Atmel
64-bit version generates binaries with 4 bytes more data than the same
32-bit version, due to padding a one-character variable to 8 bytes
instead of 4, and that would also move the rest of the data by 4 bytes
and change which variable get corrupted.
- something else that I haven't though of...

> I would like to help more if you need, be programming, giving ideas or reporting bugs!

Thanks. Do let me know if -rc1 or -rc2 show any bugs. I'll run some
more tests to see if I can find out what the deciding factor is. Of
course, I'm hoping that the cause is that my compiler works and
Atmel's doesn't but that's just my big-headedness ;-)

   M
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[Martin Guy]

OK, I've tried all builds:
- elua-trunk-0.9 on 32-bit Ubuntu
- elua-trunk-0.9 on 64-bit Debian
- elua-webbuild-0.8 on simplemachines web builder
- elua-master-0.9 on simplemachines web builder
with compilers
- Atmel GNU Toolchain 3.4.1
- my ct-ng one

In all cases, with Atmel 3.4.1 toolchain they hang on net.lookup()
All the ct-ng builds work.

In the meantime, Atmel have released their GNU Toolchain 3.4.2, which
my ct-ng is based on, and all versions work OK with that too.

So it looks like Atmel 3.4.1 was generating wrong code, fixed in their
3.4.2 release.

I've updated the compiler that builder.simplemachines.it uses to Atmel
3.4.2, so the firmware generated by the web builder should work now.

Vagner, I suggest you update your web builder's avr32 toolchain to
Atmel 3.4.2 also.

And thanks to Gabriel for insisting about the bug! :)

(BTW, Gabriel, even with a working system it can happen that
net.lookup() returns 0, which seems to mean "I don't know", but trying
it again succeeds.
You might like to test for 0 and retry a few times if it does this)

   M
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev