Puppy Linux Discussion Forum

[Jesse]

Hi Barry,

Are there any clues about what has gone wrong, if you run it with the -d option?
Sorry I don't have a modem here to test with.

What do you think is best course of action, debug and fix 1.1f, or go back to 1.0 and patch things one at a time?

Jesse

[BarryK]

zygo: no, only genuine hardware modems.

Jesse: -d option returns this:

Probing serial ports...
ensure_dev_node_present on /dev/ttyS0 4 64
Found [/dev/ttyS0]
ensure_dev_node_present on /dev/ttyS1 4 65
ensure_dev_node_present on /dev/ttyS2 4 66
ensure_dev_node_present on /dev/ttyS3 4 67
ensure_dev_node_present on /dev/ttyS4 4 68

...which is interesting, at least it recognises something is there!
Does that help to narrow-down the problem?

[Jesse]

Hi Barry,

Yep, the problem is certainly narrowed down, I just had a quick look at the code and spotted the logic flaw in it, and I looked at the original code, and scratching my head as to why that worked, the original isn't gauranteed to work, but that is determined at compile time, it needs a compiler optimization to be gauranteed for the program to work, which is what probably confused me in the first place.
I'll hopefully post an update soon.

Jesse

[BarryK]

....how's the big-finding going?
I want to wait until this is fixed before ploading 2.17alpha.

[BarryK]

Okay, I got it working.

I realised that Jesse has a problem not having a serial modem to test with, so got stuck into it myself.

It's a very bad hack though. I went back to the original version, which tests serial modems only. I then turned on the mouse detection bits, which waswhat Dougal did for v1.1, but it didn't detect my ps/2 mouse -- I then tried Dougal's v1.1, same problem.

So, I grabbed Jesse's 'mouse.c' and hacked it very badly, but it now works (don't cringe too much when you look at it, Jesse )

I'm only using puppyserialdetect in Puppy to detect modems and mice, not keyboards, as I'm still using some code that pakt works out sometime ago for usb keyboard detection.

I'm called it version 1.2...

[Jesse]

Hi Barry,

I've been a bit frantic at work the last few weeks, yep, quite unusual for a telco. I havn't really had an oppertunity to sit down with puppyserialdetect and have a fully functional brain, but no other plans for saturday and sunday, so I should have the oppertunity to hae a go

Jesse

[zygo]

I don't understand. How should this be tested?
Like this:

[Pence]

sh-3.00# ./puppyserialdetect
Type:modem|Port:/dev/ttyS1|Speed:230400
Type:PS2-mouse|Port:/dev/input/mice
sh-3.00#
I'm not sure if I did this proper.I extracted the download then opened the folder and then I right clicked the wheel named Puppyserialdetect, then clicked Window -> Terminal Here, Typed ./puppyserialdetect then pressed enter.
The modem lights started flashing for a second or two and I got that result.

[Jesse]

Hello,

Pence, Zygo, thanks for the test report, looks like Barrys 1.2 is good.

Barry, I havn't looked so no cringing here , but looks like it works.


I've updated my edition, hopefully patched the bugs etc, and have added some more debug lines in case its still not working.
Also added a -u command line option to list serial ports found that have nothing attached.
If anyone would like to test this one, let us know how it goes.

[zygo]

1.1g gives the same results as 1.2 including the change when the path is included on the command line. Jesse can you explain this?

[Pence]

I have two serial ports on the back of my computer.I connected two modems and turned them on and got this;
sh-3.00# ./puppyserialdetect
Type:PS2-mouse|Port:/dev/input/mice
Type:modem|Port:/dev/ttyS0|Speed:230400
Type:modem|Port:/dev/ttyS1|Speed:230400
sh-3.00#

[Jesse]

Hi,

Pence, all looks good! thanks for trying it out.

Zygo, 1.2 and 1.1g, are both good versions it seems. 1.1g has certainly been bug fixed now that it is reporting your serial modem.
Theoretically, if there are no bugs in either 1.2 or 1.1g, then certainly the output from the programs will be the identical, this is good!

Jesse

[zygo]

Jesse,

I was unclear. Sorry.

Why does invoking a program without it's path (my first code box) produce different results from when its path is included (my second code box)?

[Jesse]

Hi Zygo,

The short answer is, because you are running a different program.

When you type in the name of an executable in a command window (without a path prefix), your command line program (bash) has some standard rules about how it goes about finding the named program. It looks in the directories listed in the PATH environment variable, in order, if it finds the named program, bash executes it.
The PATH environment variable does not include the current directory, and its considered a security violation, by most linux/unix people.

If you provide the pathname to the executable, then bash does not need to scan the PATH variable hunting for the program, it just executes it.

I hope that fills in the blanks for you?

Jesse

[zygo]

Jesse,

Thanks for clearing that up. I am happy that I was executing 2 programs. This behaviour is familiar from my dos days. I knew linux had a path variable but as you have seen when tested I didn't put the two together.