| 1 |
<?xml version="1.0" encoding="ISO-8859-1"?> |
| 2 |
<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN" "http://www.oasis-open.org/docbook/xml/4.4/docbookx.dtd"> |
| 3 |
|
| 4 |
<chapter id="backend-spawn"> |
| 5 |
<title>Spawned Backends</title> |
| 6 |
<sect1 id="backends-spawn-intro"> |
| 7 |
<title>Introduction</title> |
| 8 |
<para> |
| 9 |
If you do not have a C or C++ language binding, PackageKit executes |
| 10 |
helper scripts written in pretty much any language. |
| 11 |
It then watches the standard out and standard error and parses the |
| 12 |
output into compiled backend commands. |
| 13 |
This means a python library can be interfaced easily with a C backend. |
| 14 |
</para> |
| 15 |
<para> |
| 16 |
Even when using helpers, a compiled backend stub is still used for |
| 17 |
two reasons: |
| 18 |
</para> |
| 19 |
<itemizedlist> |
| 20 |
<listitem> |
| 21 |
<para> |
| 22 |
It is still needed for the dlopen internally in PackageKit. |
| 23 |
</para> |
| 24 |
</listitem> |
| 25 |
<listitem> |
| 26 |
<para> |
| 27 |
You can add cleverness in the C backend that you might not want to |
| 28 |
do in the scripted backend, for example using a hash table in C |
| 29 |
rather than checking all the names in awk. |
| 30 |
</para> |
| 31 |
</listitem> |
| 32 |
</itemizedlist> |
| 33 |
<para> |
| 34 |
Backends are typically open-programmable, which means we can define a |
| 35 |
standard for what goes on stdin and stdout to try and maximise |
| 36 |
the common code between the backends. |
| 37 |
</para> |
| 38 |
<para> |
| 39 |
If you are unable to write scripts that conform to these specifications |
| 40 |
then just launch a PkSpawn object in the compiled helper with stdout |
| 41 |
callbacks and then try to do screenscraping in the backend. |
| 42 |
</para> |
| 43 |
<para> |
| 44 |
Backends scripts are run with arguments and data is sent to standard out |
| 45 |
and standard error asynchronously so that PackageKit can proxy this to D-Bus. |
| 46 |
A method has command line arguments separated with tabs, and data is also |
| 47 |
separated with tabs. |
| 48 |
</para> |
| 49 |
<para> |
| 50 |
It is important to flush the standard output after each output, else |
| 51 |
Linux will helpfully buffer the output into more efficient size chunks. |
| 52 |
If you do not flush, then there will be a long IPC delay. |
| 53 |
Flushing can be achieved in C using <literal>fflush</literal> or in python |
| 54 |
using <literal>sys.stdout.flush()</literal>. |
| 55 |
</para> |
| 56 |
<para> |
| 57 |
The "dispatcher" mode is where a command is used to startup the |
| 58 |
backend, for instance <literal>yumBackend.py search-name none power</literal> |
| 59 |
and then the backend then sits and waits for more standard input. |
| 60 |
Further operations can be done on the loaded backend sending commands |
| 61 |
to stdin, e.g. <literal>search-name none power</literal>. |
| 62 |
If there are no more operations after a preset time (default 5 seconds) |
| 63 |
then the backend is sent <literal>exit</literal> over stdin, and the |
| 64 |
backend terminates. |
| 65 |
The daemon will ensure the operations are serialised, and that backends |
| 66 |
not sending <literal>finished</literal> are cleaned up properly. |
| 67 |
</para> |
| 68 |
<para> |
| 69 |
The dispatcher mode does not have to implemented in python; any |
| 70 |
language that can read from stdin can block and be used in this way. |
| 71 |
</para> |
| 72 |
</sect1> |
| 73 |
|
| 74 |
</chapter> |