rabbit-semi : Messages : 39547-39576 of 39576

All,

There are openings  for the following Jobs at my present Employer( LATENS ).

JTE-HYD-09
STE-HYD-09
SSE-WEB-11-09
SE-WEB-11-09

Please read the requirements  posted on  Latens internet at:
http://www.latens.com/about/careers.html

Please refer your friends & colleagues and make sure they mention my name (Syed
Yaqoob Ali) when they send their profiles to jobs@latens.



Happy Hunting,
Ali..

#39548

From: "virgo.boyz" <virgo.boyz@...>
Date: Mon Nov 16, 2009 5:34 am
Subject: Radio Button on Rabbit WEB

virgo.boyz

Offline

Hi,

i'm newbie on rabbit.

i want to control DS2 LED over ZHTML with radio button, but failed

any one can help me how to do this?

thanks,

#39549

From: Kenny Millar <kennymillar@...>
Date: Mon Nov 16, 2009 9:38 am
Subject: RE: Radio Button on Rabbit WEB

kennymillar2000

Offline

We’re going to need more details!

What module are you using?

What have you tried so far?

What went wrong / didn’t work when you ‘failed’ ?

Have you looked at the rabbit web samples? One of them does exactly what you want.

-K

From: rabbit-semi@yahoogroups.com [mailto:rabbit-semi@yahoogroups.com] On Behalf Of virgo.boyz
Sent: 16 November 2009 05:34
To: rabbit-semi@yahoogroups.com
Subject: [rabbit-semi] Radio Button on Rabbit WEB

Hi,

i'm newbie on rabbit.

i want to control DS2 LED over ZHTML with radio button, but failed

any one can help me how to do this?

thanks,

#39550

From: "virgo.boyz" <virgo.boyz@...>
Date: Mon Nov 16, 2009 6:10 pm
Subject: Re: Radio Button on Rabbit WEB

virgo.boyz

Offline

Hi Millar,

Yea, sorry I forgot mention the details,

I using RCM 4300 Development board,
So far I have tried and follow the sample from rabbit web documentation
And I have solved my problem (please see my code below: ZHTML CODE and C-CODE),
thanks for your respond.

But I have another problem,

Let say, the "led1" state on board is "ON" (bit PB.2=0) but on the web page (for
the first time access) there is no radio button checked, of course this is wrong
information for user.
So, I want the "form" lookup hardware condition first and then set which one
radio button should be checked.

Sorry if my English not well, can you help me how to do this?

Thanks.


--------------------------------------------------------------------------------\
--
ZHTML CODE
--------------------------------------------------------------------------------\
--
<html>
<head>
<title>LED 1</title>
</head>
<body>
<form id="form1" name="form1" method="post" action="">
<label>ON
	 <input name="led1" type="radio" value="0" onclick="submit()"
	 <?z if ($lamp1==0) { ?>
	 CHECKED
	 <?z } ?>
	 />
</label>
<label>
	 <input name="led1" type="radio" value="1" onclick="submit()"
	 <?z if ($lamp1==1) { ?>
       CHECKED
       <?z } ?>
	 />
OFF</label>
</form>
</body>
</html>
--------------------------------------------------------------------------------\
--


=================================================================
C CODE
=================================================================
.
.
.
.
int led1;
#web led1

void main(void)
{
	 brdInit();
   	 sock_init();               // initialize TCP/IP stack
   	 http_init();               // initialize web server
   	 http_set_path("/", "index.html");
   	 tcp_reserveport(80);


    while(1)
    {
       http_handler();
       BitWrPortI(PBDR, &PBDRShadow, led1, 2);
    }
}

==================================================================




--- In rabbit-semi@yahoogroups.com, Kenny Millar <kennymillar@...> wrote:
>
> We're going to need more details!
>
>
>
> What module are you using?
>
> What have you tried so far?
>
> What went wrong / didn't work when you 'failed' ?
>
> Have you looked at the rabbit web samples? One of them does exactly what you
> want.
>
> -K
>
>
>
> From: rabbit-semi@yahoogroups.com [mailto:rabbit-semi@yahoogroups.com] On
> Behalf Of virgo.boyz
> Sent: 16 November 2009 05:34
> To: rabbit-semi@yahoogroups.com
> Subject: [rabbit-semi] Radio Button on Rabbit WEB
>
>
>
>
>
> Hi,
>
> i'm newbie on rabbit.
>
> i want to control DS2 LED over ZHTML with radio button, but failed
>
> any one can help me how to do this?
>
> thanks,
>

#39551

From: "lwoodzw" <lwoodzw@...>
Date: Mon Nov 16, 2009 8:01 pm
Subject: Re: RCM5700 ethernet - FIXED - Take note!

lwoodzw

Offline

On reset, the PHY chip used in the RCM57xx family samples the LED signals as
inputs to set the internal PHY address, then switches them to outputs.  That is
why the documented design for the RCM57xx Ethernet circuit (Tech Note 266) has
the 1K resistors on the ACT and LINK signals, and why we recommend that
customers follow our designs :v)  All you really need are the appropriate
pullup/downs on those two lines if you don't want actual LEDs.

That said, we _really_ should have documented that particular PHY functionality
since it is quite non-intuitive.  I've submitted documentation updates for the
RCM5700 User's Manual and TN266 already.

-Lynn

--- In rabbit-semi@yahoogroups.com, "eilidhs_daddy" <eilidhsdad@...> wrote:
>
> I just spent 10 days and a few hundred quid trying to figure out why my
Ethernet interface wouldn't work on a board I designed for the RCM5700
>
> Turns out that you HAVE to implement the ACT and LINK leds before the
interface will work - no idea why.
>
> The ACT led must be connected from the ACT pin via 1K resistor then LED to
+3v3
>
> The LINK led must be connected from the LINK pin via 1K resistor then LED to
ground.
>
> Any other combination and the interface won't come up!
>

#39552

From: "rmcdonld" <rickie_mcdonald@...>
Date: Mon Nov 16, 2009 9:35 pm
Subject: Absolute code placement

rmcdonld

Offline

I would like to know how to place an xmem function with an absolute address? I'm
using DC 10.54 and an RCM5700. I tried #xcodorg but it gives:
line  353 : ERROR BOOTMANAGER.C   : "?�?�???�?�????" is deprecated.  Please
use "?`?`?`??" instead.
Don't have a clue what to use instead. lol

Thanks in advance.

#39553

From: Kenny Millar <kennymillar@...>
Date: Tue Nov 17, 2009 9:39 am
Subject: RE: Re: RCM5700 ethernet - FIXED - Take note!

kennymillar2000

Offline

Thanks so much for doing that – it will save future designers some headaches!

From: rabbit-semi@yahoogroups.com [mailto:rabbit-semi@yahoogroups.com] On Behalf Of lwoodzw
Sent: 16 November 2009 20:02
To: rabbit-semi@yahoogroups.com
Subject: [rabbit-semi] Re: RCM5700 ethernet - FIXED - Take note!

On reset, the PHY chip used in the RCM57xx family samples the LED signals as inputs to set the internal PHY address, then switches them to outputs. That is why the documented design for the RCM57xx Ethernet circuit (Tech Note 266) has the 1K resistors on the ACT and LINK signals, and why we recommend that customers follow our designs :v) All you really need are the appropriate pullup/downs on those two lines if you don't want actual LEDs.

That said, we _really_ should have documented that particular PHY functionality since it is quite non-intuitive. I've submitted documentation updates for the RCM5700 User's Manual and TN266 already.

-Lynn

--- In rabbit-semi@yahoogroups.com, "eilidhs_daddy" <eilidhsdad@...> wrote:
>
> I just spent 10 days and a few hundred quid trying to figure out why my Ethernet interface wouldn't work on a board I designed for the RCM5700
>
> Turns out that you HAVE to implement the ACT and LINK leds before the interface will work - no idea why.
>
> The ACT led must be connected from the ACT pin via 1K resistor then LED to +3v3
>
> The LINK led must be connected from the LINK pin via 1K resistor then LED to ground.
>
> Any other combination and the interface won't come up!
>

#39554

From: "Jeffrey" <jeff_shapiro@...>
Date: Tue Nov 17, 2009 10:06 pm
Subject: RCM5700 Interface Board Schematic - RJ45 TX+/- Flipped?

jeff_shapiro

Offline

Hi,

   I'm designing a carrier board for a RCM5750 and am working from the RCM5700
Interface Board Schematic from Rabbit.  I just noticed that the TX+ and TX-
signals from the RJ45 jack (with integrated magnetics) go to the TX- and TX+
(flipped) on the Rabbit module in the schematic.  I've been trying to examine
the actual interface board but since it's surface mount with teeny tiny traces
and small value resistors I'm not having much luck with either a multimeter or
my eyes (with magnification).  The RX+ and RX- go to their respective ports on
the module (plus to plus and minus to minus).  Just seems odd that the RX plus
and minus match up but the TX plus and minus do not.  Maybe since the signals
are going through transformers (within the jack) it doesn't matter?

   Thanks!

  -- Jeff

#39555

From: "eilidhs_daddy" <eilidhsdad@...>
Date: Wed Nov 18, 2009 10:04 am
Subject: Re: RCM5700 Interface Board Schematic - RJ45 TX+/- Flipped?

eilidhs_daddy

Offline

I noticed this too when recently laying a board for RCM5700
Other documents from Rabbit show TX+ and TX- going to the pins you'd expect.

I did some experimenting and discovered that it's probably a mistake in the
drawing. In my design I connected TX+ to TX+ and TX- to TX- and it works
perfectly.

Please remember that you need to pull the ACT signalto +3v3 via 1K (and  led if
required) and the LINK light down to gnd via 1K (+led) or the PHY will come up
with the wrong address and wont work.

-K

--- In rabbit-semi@yahoogroups.com, "Jeffrey" <jeff_shapiro@...> wrote:
>
> Hi,
>
>   I'm designing a carrier board for a RCM5750 and am working from the RCM5700
Interface Board Schematic from Rabbit.  I just noticed that the TX+ and TX-
signals from the RJ45 jack (with integrated magnetics) go to the TX- and TX+
(flipped) on the Rabbit module in the schematic.  I've been trying to examine
the actual interface board but since it's surface mount with teeny tiny traces
and small value resistors I'm not having much luck with either a multimeter or
my eyes (with magnification).  The RX+ and RX- go to their respective ports on
the module (plus to plus and minus to minus).  Just seems odd that the RX plus
and minus match up but the TX plus and minus do not.  Maybe since the signals
are going through transformers (within the jack) it doesn't matter?
>
>   Thanks!
>
>  -- Jeff
>

#39556

From: "Derek J. Lassen" <xl198cv56@...>
Date: Wed Nov 18, 2009 10:34 am
Subject: (corrected) Re: RCM5700 Interface Board Schematic - RJ45 TX+/- Flipped?

xl198cv56

Offline

Sorry - that should read:
Many PHY implementations can do auto MDI/MDIX to provide crossover so it doesn't need a crossover cable.

(s) Derek

At 22:06 11/17/2009 +0000, you wrote:

Hi,

I'm designing a carrier board for a RCM5750 and am working from the RCM5700 Interface Board Schematic from Rabbit. I just noticed that the TX+ and TX- signals from the RJ45 jack (with integrated magnetics) go to the TX- and TX+ (flipped) on the Rabbit module in the schematic. I've been trying to examine the actual interface board but since it's surface mount with teeny tiny traces and small value resistors I'm not having much luck with either a multimeter or my eyes (with magnification). The RX+ and RX- go to their respective ports on the module (plus to plus and minus to minus). Just seems odd that the RX plus and minus match up but the TX plus and minus do not. Maybe since the signals are going through transformers (within the jack) it doesn't matter?

Thanks!

-- Jeff

#39557

From: "Stan" <stanley_hsu@...>
Date: Wed Nov 18, 2009 12:21 pm
Subject: BL2100 LCD Header for use as Analog Input

stanley_hsu

Offline

Hi.

I would like to know if someone has ever made any board that could be used to
plug into what is the LCD header on the BL2100 for use as a 16 bit AD.  If so,
how many AD does it have (best if there is 2).  I would like to see if I can buy
this.

If not, depending on cost, is anyone interested in seeing if this is possible
for design etc?

Thanks.

Stan the Man

#39558

From: "tijesmurf" <tijesmurf@...>
Date: Wed Nov 18, 2009 4:08 pm
Subject: Redirecting echoed data in debug window to decision statement

tijesmurf

Offline

Hi!

I have data being sent from a VB.net application via bluetooth to an RCM3100
with an eb506 module. The data is being echoed to the debug port, so I know that
the correct data is being received. What I would now like to do is, if the data
is correct in the debug window, redirect the data to be used in a decision
statement.

For example:
If the data in the debug port reads "ON", activate some pin high or low?

Can anyone assist with this?

Stefan

#39559

From: kevin asato <kc6pob@...>
Date: Thu Nov 19, 2009 5:55 am
Subject: Re: Re: RCM5700 Interface Board Schematic - RJ45 TX+/- Flipped?

kc6pob

Offline

with transformer coupling, flipping + and - leads are not an issue. I have done this on a few occaisions with T1 signals. The problem would result when using an older device which uses bi-polar (transistor) line drivers in which the polarity did make a difference.

kevin

kc6pob

--- On Wed, 11/18/09, eilidhs_daddy <eilidhsdad@...> wrote:

From: eilidhs_daddy <eilidhsdad@...>
Subject: [rabbit-semi] Re: RCM5700 Interface Board Schematic - RJ45 TX+/- Flipped?
To: rabbit-semi@yahoogroups.com
Date: Wednesday, November 18, 2009, 2:04 AM

I noticed this too when recently laying a board for RCM5700
Other documents from Rabbit show TX+ and TX- going to the pins you'd expect.

I did some experimenting and discovered that it's probably a mistake in the drawing. In my design I connected TX+ to TX+ and TX- to TX- and it works perfectly.

Please remember that you need to pull the ACT signalto +3v3 via 1K (and led if required) and the LINK light down to gnd via 1K (+led) or the PHY will come up with the wrong address and wont work.

-K

--- In rabbit-semi@ yahoogroups. com, "Jeffrey" <jeff_shapiro@ ...> wrote:
>
> Hi,
>
> I'm designing a carrier board for a RCM5750 and am working from the RCM5700 Interface Board Schematic from Rabbit. I just noticed that the TX+ and TX- signals from the RJ45 jack (with integrated magnetics) go to the TX- and TX+ (flipped) on the Rabbit module in the schematic. I've been trying to examine the actual interface board but since it's surface mount with teeny tiny traces and small value resistors I'm not having much luck with either a multimeter or my eyes (with magnification) . The RX+ and RX- go to their respective ports on the module (plus to plus and minus to minus). Just seems odd that the RX plus and minus match up but the TX plus and minus do not. Maybe since the signals are going through transformers (within the jack) it doesn't matter?
>
> Thanks!
>
> -- Jeff
>

#39560

From: "yair24" <yairb@...>
Date: Sat Nov 21, 2009 9:47 am
Subject: http_flashspec[] authentication. upgrading DC version from 7.25 to 9.62

yair24

Offline

Hello,

I have a code that I need to modify.

the code runs on a rabbit 2200 module with ethernet.
it is a website that read some data from the board and transmit it to the
website.


I have there a const HttpSpec array called http_flashspec[] that is initialized
with many lines such as the following:

const HttpSpec http_flashspec[] =
{
  {HTTPSPEC_FILE,"/",index_html,  NULL, 0, NULL, NULL},
  {HTTPSPEC_FILE,"/index.html",index_html,  NULL, 0, NULL, NULL},
  {HTTPSPEC_FILE,"/SetIPpar.shtml",SetIPPar_shtml, NULL, 0, NULL, spr}
.
.
.
};

and more lines.

the problem I have is with the last line that has the spr var at at the end.

the spr is a variable of type "ServerPermissions", that should popup a
authentication popup before opening the "SetIpPar" page.
it used to be a "httprealm" variable because this software was compiled in a
7.25 dynamic C version. I was asked to modify the software so it can be compiled
in a 9.62 dynamic C compiler due to a problem that occured in that software when
they compiled it in the 7.25 ver.

so I saw that I need to change the httprealm to a ServerPermission in order to
adjust the software to the 9.62 compiler.

when I try to compile this I get an error:
"constant initializer expected"
and it points to the NULL before the spr var in that line.


I must mention that this is the first time I deal with a tcpip rabbit. (I have
many years of experience with rabbit 2020 with no TCPIP on it.)

can anybody see what is my mistake here? or give me a link to read more about
how to make the modification I want to do?

thanks in advanced.

#39561

From: "Alan Matheson" <alan_matheson@...>
Date: Mon Nov 23, 2009 11:46 pm
Subject: Windows 7

alan_matheson

Offline

Windows 7 doesn't like Dynamic C 9.24 and reports it as incompatible even when I try running in earlier modes. It seems to work OK for a while then does things like take a very long time to start up, 30 seconds or so, eventually producing the blue screen of death after a few slow startups. Is there any fix for this problem or am I going to have to use an older computer?

I should say that apart from this I am delighted with Windows 7.

Alan Matheson

Reply | Forward | Messages in this Topic (8)

#39562

From: "seecwriter" <seecwriter@...>
Date: Tue Nov 24, 2009 1:44 am
Subject: Debug suggestion

seecwriter

Offline

I have a procedure that collects some data in a loop, and then sends that data
out the selected port. The data is standard ASCII with a leading STX character
and a closing ETX character. The port is either one of the serial ports or the
ethernet port. If the ethernet port is used, a CR/LF pair is added after the
ETX.
I haven't experienced any problems when using the serial ports. But when I use
the ethernet port, I don't get the full message. If I run the code under the
debugger, it won't fail. If I add a printf statement to send debug data out
another serial port, the problem won't manifest it self either. If I just format
a string using sprintf, and then throw the string away, the procedure works
fine. The ethernet connection is TCP/IP.
Does anyone have any other ideas on how I might troubleshoot this?
I'm using DC 9.25.

#39563

From: "Francesco Costella" <costellaf@...>
Date: Tue Nov 24, 2009 6:32 am
Subject: RE: Windows 7

checco_bt

Offline

http://www.rabbit.com/documentation/docs/refs/TN257/TN257.htm

Hi Alan,

Check this link:

Francesco

From: rabbit-semi@yahoogroups.com [mailto:rabbit-semi@yahoogroups.com] On Behalf Of Alan Matheson
Sent: 24 November 2009 00:46
To: rabbit-semi@yahoogroups.com
Subject: [rabbit-semi] Windows 7

I should say that apart from this I am delighted with Windows 7.

Alan Matheson

Reply | Forward | Messages in this Topic (8)

#39564

From: "monte_dalrymple" <monted@...>
Date: Tue Nov 24, 2009 3:34 pm
Subject: The Evolution of Rabbits

monte_dalrymple

Offline

If anyone is interested, I have an article entitled "The Evolution
of Rabbits" in the December digital-plus edition of Circuit Cellar
magazine. It talks about the process and challenges of designing
five generatons of Rabbit microprocessors.

Monte Dalrymple

#39565

From: "seecwriter" <seecwriter@...>
Date: Tue Nov 24, 2009 7:38 pm
Subject: Re: Debug suggestion

seecwriter

Offline

I've been sniffing the network with wireshark to see if I can get a clue as to
what is happening. Perhaps someone can help decipher what I'm seeing.

I first want to give a little more detail on how data is sent.

The Rabbit logs data in 64-byte entry strings. When requested, 10-entries are
returned. But I don't create a 640-byte buffer to store the entries in. I have
one 80-byte buffer that I load with one entry and send, then repeat with the
next entry, etc., until all 10 are sent. This works great for serial, but not
for ethernet.

This is my ethernet routine that is called for each entry sent:

void SendEnet( tcp_socket *s, char *p, int len) {

   sock_flushnext(s);
   sock_awrite(s,p,len);
   tcp_tick(s);
}

On wireshark I see the command from the host to the Rabbit requesting the data.

The next packet is a response from the Rabbit with the data header.

The next packet is a response from the Rabbit with 8 of the 10 entries. I
expected one entry per packet. That may or may not be relevant.

The next packet is from the host to the rabbit. The "info" section for this
packet is: anet-1 > cisco-sccp [ACK] Seq=40 Ack=1953 Win=65049 Len=0.
I'm assuming this is an acknowledge from the host.

The next packet appears to be marked as a duplicate of a previous packet. The
"info" for this packet is: [TCP Dup ACK 3136#1] anet1 [ACK] Seq=1953 Ack=40
Win=512 Len=0

At this point there is no further communication, and I've only received 8 of the
10 entries.

Are there any clues here?

--- In rabbit-semi@yahoogroups.com, "seecwriter" <seecwriter@...> wrote:
>
> I have a procedure that collects some data in a loop, and then sends that data
out the selected port. The data is standard ASCII with a leading STX character
and a closing ETX character. The port is either one of the serial ports or the
ethernet port. If the ethernet port is used, a CR/LF pair is added after the
ETX.
> I haven't experienced any problems when using the serial ports. But when I use
the ethernet port, I don't get the full message. If I run the code under the
debugger, it won't fail. If I add a printf statement to send debug data out
another serial port, the problem won't manifest it self either. If I just format
a string using sprintf, and then throw the string away, the procedure works
fine. The ethernet connection is TCP/IP.
> Does anyone have any other ideas on how I might troubleshoot this?
> I'm using DC 9.25.
>

#39566

From: Scott Henion <shenion@...>
Date: Tue Nov 24, 2009 8:43 pm
Subject: Re: Re: Debug suggestion

shdesigns2003

Offline

seecwriter wrote:
> I've been sniffing the network with wireshark to see if I can get a clue as to
what is happening. Perhaps someone can help decipher what I'm seeing.
>
> I first want to give a little more detail on how data is sent.
>
> The Rabbit logs data in 64-byte entry strings. When requested, 10-entries are
returned. But I don't create a 640-byte buffer to store the entries in. I have
one 80-byte buffer that I load with one entry and send, then repeat with the
next entry, etc., until all 10 are sent. This works great for serial, but not
for ethernet.
>
> This is my ethernet routine that is called for each entry sent:
>
> void SendEnet( tcp_socket *s, char *p, int len) {
>
>   sock_flushnext(s);
>   sock_awrite(s,p,len);
>   tcp_tick(s);
> }
>
> On wireshark I see the command from the host to the Rabbit requesting the
data.
>
> The next packet is a response from the Rabbit with the data header.
>
> The next packet is a response from the Rabbit with 8 of the 10 entries. I
expected one entry per packet. That may or may not be relevant.
>
> The next packet is from the host to the rabbit. The "info" section for this
packet is: anet-1 > cisco-sccp [ACK] Seq=40 Ack=1953 Win=65049 Len=0.
> I'm assuming this is an acknowledge from the host.
>
> The next packet appears to be marked as a duplicate of a previous packet. The
"info" for this packet is: [TCP Dup ACK 3136#1] anet1 [ACK] Seq=1953 Ack=40
Win=512 Len=0
>
> At this point there is no further communication, and I've only received 8 of
the 10 entries.
>
> Are there any clues here?
>
>
>

What is your code doing at this point. Is it calling tcp_tick() so the
rest can be sent out after the Nagle timeout?


--
------------------------------------------
| Scott G. Henion| shenion@... |
|   Consultant   |   Stone Mountain, GA  |
|   SHDesigns http://www.shdesigns.org   |
------------------------------------------

today's fortune
Steinbach's Guideline for Systems Programming:
	 Never test for an error condition you don't know how to handle.

#39567

From: "seecwriter" <seecwriter@...>
Date: Tue Nov 24, 2009 8:54 pm
Subject: Re: Debug suggestion

seecwriter

Offline

>
>What is your code doing at this point. Is it calling tcp_tick() so >the rest
can be sent out after the Nagle timeout?

Yes. After calling SendEnet 11 times (1 header, 10 entries) it goes back to the
main loop where it calls tcp_tick, among other things.

Steve

#39568

From: Scott Henion <shenion@...>
Date: Tue Nov 24, 2009 9:02 pm
Subject: Re: Re: Debug suggestion

shdesigns2003

Offline

seecwriter wrote:

What is your code doing at this point. Is it calling tcp_tick() so >the rest can be sent out after the Nagle timeout?


Yes. After calling SendEnet 11 times (1 header, 10 entries) it goes back to the main loop where it calls tcp_tick, among other things.

What are you uing to write to the socket?
Are you checking the return value of the write function?

-- ------------------------------------------
| Scott G. Henion| shenion@... |
| Consultant | Stone Mountain, GA |
| SHDesigns http://www.shdesigns.org |
------------------------------------------
today's fortune I came out of twelve years of college and I didn't even know how to sew.
All I could do was account -- I couldn't even account for myself.
-- Firesign Theatre

#39569

From: "seecwriter" <seecwriter@...>
Date: Tue Nov 24, 2009 9:51 pm
Subject: Re: Debug suggestion

seecwriter

Offline

I am not checking the return value of the write function. I know that is
verbotin, but I'm not sure what my program would do if the write failed?

If I check the return value, and then send the result out the debug port, I need
to have debug enabled. When debug is enabled, it doesn't fail. And looking at
the ethernet traffic with wireshark with debug enabled, I get a lot of packets
that have one or two entries per packet. There are no packets have more than two
entries.
Perhaps this suggests that because debug causes slower code, the enet buffers
have time to be freed up before the next write. I have TCP_BUF_SIZE set to 1024,
and the total amount of data being sent is 640.

Steve

--- In rabbit-semi@yahoogroups.com, Scott Henion <shenion@...> wrote:
>
> seecwriter wrote:
> >> What is your code doing at this point. Is it calling tcp_tick() so >the
rest can be sent out after the Nagle timeout?
> >>
> >
> > Yes. After calling SendEnet 11 times (1 header, 10 entries) it goes back to
the main loop where it calls tcp_tick, among other things.
> >
> >
>
> What are you uing to write to the socket?
> Are you checking the return value of the write function?
>
>
> --
> ------------------------------------------
> | Scott G. Henion| shenion@... |
> |   Consultant   |   Stone Mountain, GA  |
> |   SHDesigns http://www.shdesigns.org   |
> ------------------------------------------
>
> today's fortune
> I came out of twelve years of college and I didn't even know how to sew.
> All I could do was account -- I couldn't even account for myself.
> 	 -- Firesign Theatre
>

#39570

From: "seecwriter" <seecwriter@...>
Date: Tue Nov 24, 2009 10:40 pm
Subject: Re: Debug suggestion

seecwriter

Offline

Well, I figured out how I can use the return value from the write. One of the
return values indicates that there is not a buffer large enough to hold the data
you're trying to send. So I changed my write function to test for that return
return value. If I get it, I call tcp_tick and try again. So I loop there until
the data gets sent, as long as I get that one return code (0).

My brief testing so far works. I'll keep testing for awhile to be sure, but it
looks like I may have stumbled across the solution with your help. Thanks Scott!

Steve


--- In rabbit-semi@yahoogroups.com, "seecwriter" <seecwriter@...> wrote:
>
> I am not checking the return value of the write function. I know that is
verbotin, but I'm not sure what my program would do if the write failed?
>
> If I check the return value, and then send the result out the debug port, I
need to have debug enabled. When debug is enabled, it doesn't fail. And looking
at the ethernet traffic with wireshark with debug enabled, I get a lot of
packets that have one or two entries per packet. There are no packets have more
than two entries.
> Perhaps this suggests that because debug causes slower code, the enet buffers
have time to be freed up before the next write. I have TCP_BUF_SIZE set to 1024,
and the total amount of data being sent is 640.
>
> Steve
>
> --- In rabbit-semi@yahoogroups.com, Scott Henion <shenion@> wrote:
> >
> > seecwriter wrote:
> > >> What is your code doing at this point. Is it calling tcp_tick() so >the
rest can be sent out after the Nagle timeout?
> > >>
> > >
> > > Yes. After calling SendEnet 11 times (1 header, 10 entries) it goes back
to the main loop where it calls tcp_tick, among other things.
> > >
> > >
> >
> > What are you uing to write to the socket?
> > Are you checking the return value of the write function?
> >
> >
> > --
> > ------------------------------------------
> > | Scott G. Henion| shenion@ |
> > |   Consultant   |   Stone Mountain, GA  |
> > |   SHDesigns http://www.shdesigns.org   |
> > ------------------------------------------
> >
> > today's fortune
> > I came out of twelve years of college and I didn't even know how to sew.
> > All I could do was account -- I couldn't even account for myself.
> > 	 -- Firesign Theatre
> >
>

#39571

From: Scott Henion <shenion@...>
Date: Tue Nov 24, 2009 11:12 pm
Subject: Re: Re: Debug suggestion

shdesigns2003

Offline

seecwriter wrote:
> Well, I figured out how I can use the return value from the write. One of the
return values indicates that there is not a buffer large enough to hold the data
you're trying to send. So I changed my write function to test for that return
return value. If I get it, I call tcp_tick and try again. So I loop there until
the data gets sent, as long as I get that one return code (0).
>
> My brief testing so far works. I'll keep testing for awhile to be sure, but it
looks like I may have stumbled across the solution with your help. Thanks Scott!
>
>
You still have not said which function you were using. It sounds like
sock_fastwrite(). That will drop any data that won't fit in the outgoing
buffers. sock_write() _will_ wait and call tcp_tick() until there is a
buffer free.

sock_fastwrite() is there if you don't want the stack blocking. Useful
when you have data to send and can easily regenerate it later if it fails.

--
------------------------------------------
| Scott G. Henion| shenion@... |
|   Consultant   |   Stone Mountain, GA  |
|   SHDesigns http://www.shdesigns.org   |
------------------------------------------

today's fortune
QOTD:
	 "I'm on a seafood diet -- I see food and I eat it."

#39572

From: "seecwriter" <seecwriter@...>
Date: Tue Nov 24, 2009 11:33 pm
Subject: Re: Debug suggestion

seecwriter

Offline

Sorry, I'm using socket_awrite. I posted my write procedure in my original post.
I changed my write function to this:

void SendEnet( tcp_socket *s, char *p, int len ) {
   int result;
   do
     socket_flushnext();
     result = socket_awrite(s, p, len);
     tcp_tick(s);
   } while ( result == 0 );


The one potential flaw in my procedure is that it will lock-up
if a buffer never becomes available. Do I care? I have to think
about.

Steve



--- In rabbit-semi@yahoogroups.com, Scott Henion <shenion@...> wrote:
>
> seecwriter wrote:
> > Well, I figured out how I can use the return value from the write. One of
the return values indicates that there is not a buffer large enough to hold the
data you're trying to send. So I changed my write function to test for that
return return value. If I get it, I call tcp_tick and try again. So I loop there
until the data gets sent, as long as I get that one return code (0).
> >
> > My brief testing so far works. I'll keep testing for awhile to be sure, but
it looks like I may have stumbled across the solution with your help. Thanks
Scott!
> >
> >
> You still have not said which function you were using. It sounds like
> sock_fastwrite(). That will drop any data that won't fit in the outgoing
> buffers. sock_write() _will_ wait and call tcp_tick() until there is a
> buffer free.
>
> sock_fastwrite() is there if you don't want the stack blocking. Useful
> when you have data to send and can easily regenerate it later if it fails.
>
> --
> ------------------------------------------
> | Scott G. Henion| shenion@... |
> |   Consultant   |   Stone Mountain, GA  |
> |   SHDesigns http://www.shdesigns.org   |
> ------------------------------------------
>
> today's fortune
> QOTD:
>  "I'm on a seafood diet -- I see food and I eat it."
>

#39573

From: "Alan Matheson" <alan_matheson@...>
Date: Wed Nov 25, 2009 2:09 am
Subject: Re: Windows 7

alan_matheson

Offline

http://www.rabbit.com/documentation/docs/refs/TN257/TN257.htm

Thanks Francesco, the problem seems to have gone away after following the notes.

From: Francesco Costella

Sent: Tuesday, November 24, 2009 7:32 PM

To: rabbit-semi@yahoogroups.com

Subject: RE: [rabbit-semi] Windows 7

Hi Alan,

Check this link:

Francesco

I should say that apart from this I am delighted with Windows 7.

Alan Matheson

Reply | Forward | Messages in this Topic (8)

#39574

From: "Alan Matheson" <alan_matheson@...>
Date: Wed Nov 25, 2009 2:49 am
Subject: Whats the best approach

alan_matheson

Offline

I eventually want to set up an instrument that uses a RCM3100 to communicate via the Thuraya satellite and send files to an ftp site.

To gain some knowledge of ftp transfers I setup an RCM3700 module connected to my router via Ethernet used DHCP to set up a socket then successfully sent files via ftp to a remote ftp server. I have to confess there was more luck than skill setting up this code but it worked

Next thing I want to do is to login to a local ISP using a dial up modem then transfer files to the same ftp site. This will closely simulate the transfer via the Thuraya satellite modem. I have to simulate because I have a tiny difficulty in that the Thuraya satellite is not visible from my location!

I HAVE to use a RCM3100 connected to a dial up modem via serial B. No choice in this, its in an existing instrument. I don't have any spare I/O lines so can't do hardware handshaking.

What is the correct way to implement this transfer? Do I log into the ISP using the PPP library set up a socket then attempt to use the ftp library or am I on the wrong track. I'm an absolute novice at this sort of programming.

Alan Matheson

#39575

From: "M. Manca" <m.manca@...>
Date: Wed Nov 25, 2009 7:50 am
Subject: Re: Re: Debug suggestion

micronpn

Offline

I would use sock_fastwrite() inside a state of a finite state machine.
seecwriter ha scritto:

Sorry, I'm using socket_awrite. I posted my write procedure in my original post. I changed my write function to this:

void SendEnet( tcp_socket *s, char *p, int len ) {
int result;
do
socket_flushnext();
result = socket_awrite(s, p, len);
tcp_tick(s);
} while ( result == 0 );

The one potential flaw in my procedure is that it will lock-up
if a buffer never becomes available. Do I care? I have to think
about.

Steve

--- In rabbit-semi@yahoogroups.com, Scott Henion <shenion@...> wrote:
>
> seecwriter wrote:
> > Well, I figured out how I can use the return value from the write. One of the return values indicates that there is not a buffer large enough to hold the data you're trying to send. So I changed my write function to test for that return return value. If I get it, I call tcp_tick and try again. So I loop there until the data gets sent, as long as I get that one return code (0).
> >
> > My brief testing so far works. I'll keep testing for awhile to be sure, but it looks like I may have stumbled across the solution with your help. Thanks Scott!
> >
> >
> You still have not said which function you were using. It sounds like
> sock_fastwrite(). That will drop any data that won't fit in the outgoing
> buffers. sock_write() _will_ wait and call tcp_tick() until there is a
> buffer free.
>
> sock_fastwrite() is there if you don't want the stack blocking. Useful
> when you have data to send and can easily regenerate it later if it fails.
>
> --
> ------------------------------------------
> | Scott G. Henion| shenion@... |
> | Consultant | Stone Mountain, GA |
> | SHDesigns http://www.shdesigns.org |
> ------------------------------------------
>
> today's fortune
> QOTD:
> "I'm on a seafood diet -- I see food and I eat it."
>

Attachment: vcard [not shown]

#39576

From: "seecwriter" <seecwriter@...>
Date: Wed Nov 25, 2009 3:17 pm
Subject: Re: Debug suggestion

seecwriter

Offline

> I would use sock_fastwrite() inside a state of a finite state
> machine.

Why? What makes sock_fastwrite a better choice?


Steve