:: Re: [DNG] Init compatibility (was: …
トップ ページ
このメッセージを削除
このメッセージに返信
著者: Florian Zieboll
日付:  
To: dng
題目: Re: [DNG] Init compatibility (was: SoylentNews discussion)
On Sat, 4 Jun 2016 03:27:16 -0400
Steve Litt <slitt@???> wrote:

> On Sat, 4 Jun 2016 00:42:46 +0200
> Florian Zieboll <f.zieboll@???> wrote:
>
> > My experience with init systems is mostly limited sysV init (well,
> > also busybox and system d). That said, I wonder, what information
> > any arbitrary init system would need, that can not be delivered
> > e.g. in a simple XML file, packaged with the daemon.
>
> You never, never, NEVER want PID1 to be encumbered with an XML parser.
> Unless you're OS/x (launchd). What could *possibly* go wrong?



Hallo Steve,

you fully got me wrong here: I don't see the xml (or what ever) parser
in the init system (aka pid1) itself, but to be called by the updating
routines of the init systems, whenever an additional daemon gets
installed.

In other words: ATM, every init system comes packaged with the necessary
init scripts in their own "proprietary" format. I envision this
information extracted from the individual init scripts and put into an
init independent, "per daemon" config file, which is parsed at
installation time (not at boot time!) by a "per init system" script to
create the then static init scripts.

So I'm pointing into a similar direction as Peter Olson did earlier
today in the "SoylentNews discussion" thread:

| On Sat, 4 Jun 2016 02:35:54 -0400 (EDT)
| Peter Olson <peabo@???> wrote:
|
| It's a difficult issue, but maybe solvable by parsing common idioms
| of each init system into a description of the init process.
|
| Some of this may be automatic, but Turing says it cannot be done
| entirely by machine. It would be nice to make a lot of parsing
| automatic and only require human intervention for the difficult cases.
|
| I don't have a specific idea of how to do this.
|
| The resulting idea is to transform the parsed description into files
| for each of the kinds of init.
|
| Difficult issues could be submitted as feature requests upstream.



I saw your mail from tonight (Fri, 3 Jun 2016 18:17:34 -0400), also in
the "SoylentNews discussion" thread, which more or less overlapped with
my previous mail, so I realize that there are more challenges to solve
than telling the init system when and how to fire up which daemon...

Regards from within the new central european monsoon,

Florian