Re: Documentation of device spec list does not agree with parser code

[Roger James <roger beardandsandals co uk>]

On 18/01/16 11:21, Thomas Haller wrote:
> On Mon, 2016-01-18 at 10:54 +0000, Roger James wrote:
> > On 17/01/16 21:04, Thomas Haller wrote:
> > > On Fri, 2016-01-15 at 18:21 +0000, Roger James wrote:
> > > > Hi,
> > > >
> > > > As far as I can see the code for parsing device interface-name
> > > > lists
> > > > in
> > > > NetworkManagerUtils.c does not agree with the man page in
> > > > NetworkManager.conf.xml.in. In particular
> > > >
> > > > interface-name:IFNAME
> > > >
> > > > does not support globbing. Only
> > > >
> > > > interface-name:?IFNAME
> > > Hi Roger,
> > >
> > > where do you read "interface-name:?IFNAME". AFAIS, that is not
> > > mentioned by upstream manpage. See.
> > >
> > > It's either:
> > >
> > >   interface-name:=<LITERAL>
> > >
> > > or
> > >
> > >    interface-name:~<PATTERN>
> > >
> > > (whereas, omitting the ~ is allowed:
> > >
> > >    interface-name:<PATTERN>
> > >
> > > http://cgit.freedesktop.org/NetworkManager/NetworkManager/tree/man/
> > > NetworkManager.conf.xml.in?id=7a2a96f8ef364e46f9c99798743c11f1ea026
> > > 59d#n920
> > >
> > >
> > >
> > > > does.
> > > >
> > > > Here is the current code.
> > > I don't see anything wrong. Can you be more specific?
> > >
> > >
> > > There is also a unit-test: http://cgit.freedesktop.org/NetworkManag
> > > er/NetworkManager/tree/src/tests/test-
> > > general.c?id=7a2a96f8ef364e46f9c99798743c11f1ea02659d#n811
> > >
> > >
> > > ciao
> > > Thomas
> > Hi Thomas,
> >
> > If I follow the link above, the file I arrive at has line 920-921
> > that looks like this
> >
> >            <term>interface-name:IFNAME</term>
> >            <term>interface-name:~IFNAME</term>
> >
> > That is looks the same as the one in my tree, which has its origin
> > at
> > origin  git://anongit.freedesktop.org/NetworkManager/NetworkMan
> > ager.git (fetch)
> > origin  git://anongit.freedesktop.org/NetworkManager/NetworkMan
> > ager.git (push)
> > There is no mention of PATTERN in the file.
> Hi Roger,
>
>
> yes, this is right.Sorry, I don't understand what the issue is.
> I don't see any problem.
>
> Can you please re-state what you think is wrong?
>
>
> Thank you,
> Thomas
Hi Thomas,

Sorry about the confusion. The ? instead of the ~ in the first message 
was a typo.

To put things in context

        <varlistentry>
          <term>interface-name:IFNAME</term>
          <term>interface-name:~IFNAME</term>
          <listitem><para>Case sensitive match of interface name of the device. Simple globbing is supported 
with
             <literal>*</literal> and <literal>?</literal>. Ranges and escaping is not 
supported.</para></listitem>
        </varlistentry>
 

Produce the following statement in the man page entry:

       interface-name:IFNAME, interface-name:~IFNAME
           Case sensitive match of interface name of the device. Simple
           globbing is supported with * and ?. Ranges and escaping is not
           supported.

This implies that that interface-name:?on* should match for example mon0.

However the code fro the nm_match_spec_interface_name function in 
NetworkManagerUtils.c, only sets the match_pattern flag if it sees a ~ 
character as the first character of the interface name. This is at or 
around line 1322 of the source file.

I think that.

        <varlistentry>
          <term>interface-name:IFNAME</term>
          <listitem><para>Case sensitive match of interface name of the device. Ranges and escaping is not 
supported.</para></listitem>
        </varlistentry>
        <varlistentry>
          <term>interface-name:~IFNAME</term>
          <listitem><para>Case sensitive match of interface name of the device using globbing with
             <literal>*</literal> and <literal>?</literal>. Ranges and escaping is not 
supported.</para></listitem>
        </varlistentry>

May be a more accurate statement of what the code implements.

However I am not very familiar with this code. So I might be completely 
mistaken. I came across this whilst trying work out why my config was 
not working on Ubuntu 15.10. However the NM version there appears to be 
very old and  I suspect that interface-names in the conf  for exclusion 
do not work for a number of other reasons :-)

Roger