Added parameter '--throw-exception', which can not be specified togheter...

[mgor]

Consider the following DNA file:

$ cat test.dna 
.Hello @{name}!

It contains a reference to an undefined python variable called name. When generating the output, there's nothing:

$ ./ribosome.py test.dna 
$ echo $?
0

When adding --rna and executing the resulting code it's quite clear the problem is a variable that isn't defined:

$ ./ribosome.py --rna test.dna | python 
Traceback (most recent call last):
  File "<stdin>", line 259, in <module>
  File "<stdin>", line 213, in dot
  File "<stdin>", line 197, in add
  File "<string>", line 1, in <module>
NameError: name 'name' is not defined

To fix this, and not break backwards compatibility I added a new parameter --throw-exception (couldn't come up with a better name...) which basically works as --rna but writing the RNA code to stdout instead. When specifying the new parameter, the try-catch-exception code parts will not be added to the generated RNA code. The parameter can not be specified at the same time as --rna (mutually exclusive):

$ ./ribosome.py --throw-exception --rna test.dna 
usage: ribosome.py [-h] [-v] [--rna | --throw-exception] dna
ribosome.py: error: argument --rna: not allowed with argument --throw-exception

The new parameter, in combination with using subprocess.Popen() instead of subprocess.call() makes it possible to read stderr for any error messages and display them:

$ ./ribosome.py --throw-exception test.dna 
Traceback (most recent call last):
  File "test.rna", line 259, in <module>
    Block.dot('Hello @{name}!', locals())
  File "test.rna", line 213, in dot
    Block.add(line, bind)
  File "test.rna", line 197, in add
    val = eval(expr, bind)
  File "<string>", line 1, in <module>
NameError: name 'name' is not defined
$ echo $?
1

All "legacy" tests passed!

[eeyrw]

I do need such kind of feature. The ribosome do not output any error information when the py has problem. It's quite hard to trace the problem. I just generate rna file then run in separate for error information.

[mgor]

In my use case, I had around 20 dna files representing a section in a big configuration file. The configuration is built using make,

[mgor]

Dammit, my son manage to send before I could finish.

...make. Without this fix there wouldn't be any indication on why make couldn't build the file. It would just silently die.

[sustrik]

Uh. Sorry for the delay. I see that GitHub reports merge conflicts for this patch. Would you give it a look?