1 |
On Fri, Apr 13, 2001 at 09:20:03AM -0600, Daniel Robbins wrote: |
2 |
> > To recap, the idea was to have a C/C++ program listen on a pipe, then have |
3 |
> > interfaces (in the language of your choice) do things like |
4 |
> > "echo packagename>/thatpipe" - the backend would listen and install the package. |
5 |
> |
6 |
> One problem with this solution is that we need a robust and feature-rich two-way |
7 |
> communications stream. For example, your request to install "foo/bar-1.0" may be |
8 |
> answered by a list of dependencies that must be satisfied before installation is |
9 |
> possible, or a particular dependency that couldn't be satisfied (if it's auto-resolving |
10 |
> dependencies), or that the particular package wasn't found in the portage database. |
11 |
> |
12 |
> With such a system as you describe, the Portage response would have to be placed |
13 |
> in another file, and tying these two files into a communications stream would become |
14 |
> tricky enough to make this a less-than-elegant, solution. IMO. |
15 |
> |
16 |
|
17 |
Well, the backend program *does* have both a stdin and stdout stream, |
18 |
which would provide the two way communication stream you're talking |
19 |
about. And we could also set up a socket file if necessary. Or even |
20 |
unix sockets (but I think that would be overkill...) |
21 |
|
22 |
Pete |