Controlling an external program with Python
Paul Duffin
pduffin at hursley.ibm.com
Thu Jun 29 07:28:17 EDT 2000
Alexandre Ferrieux wrote:
>
> Paul Duffin wrote:
> >
> > Apparently there is a port of Expect for Python so why not use that
> > rather than try and reimplement it all over again.
> > Expect has over 7 years of development and centuries of testing on many
> > different platforms, do you really want to reproduce that. If so why,
> > what exactly is wrong with Expect.
>
> The same as with TkInter: it needs a full Tcl inside. Again I've got
> next to nothing against Tcl, it's just that the separation of orthogonal
> things is such a natural thing for me.
>
I agree which is why I suggested that rather than rewrite Expect in Python
which is no more *orthogonal* than Expect in Tcl why not do as I suggested
and work with Don Libes on seperating the pseudo tty control stuff into
a seperate library which both Tcl and Python can use. Everyone would
gain and noone would lose, the same could be done for Tk.
+-----------+ +----------+
| TclExpect | | PyExpect |
+-----------+ +----------+
+------------------------+
| Pseudo TTY management |
+------------------------+
The above structure is much more *orthogonal* than the following structures.
+--------TclExpect-------+ +--------PyExpect--------+
| | | |
| Pseudo TTY management | | Pseudo TTY management |
| | | |
+------------------------+ +------------------------+
+----------PyExpect----------+
| +--------TclExpect-------+ |
| | | |
| | Pseudo TTY management | |
| | | |
| +------------------------+ |
+----------------------------+
This sort of arrangement is only possible when people who are interested in
Expect functionality can transcend any language *issues*, by accepting that
the language is basically irrelevant apart from the effect in may have on
the required functionality and work together.
More information about the Python-list
mailing list