Hi Robert,
it make sense now that you explained it. Thank you.
Regards,
Martin (aka SloMo)
On 06.03.2007 23:33, Robert Rainwater wrote:
> Hi,
>
> I should be more clear about the intention of the
latest changes.
> MirandaPluginInterfaces and MirandaPluginInfoEx will
always be required
> in 0.8. In the current trunk, if you use the new
MirandaPluginInfoEx
> you *must* use the MirandaPluginInterfaces function.
>
> It may sound unreasonable at first glance, but
requiring a plugin define
> at least 1 interface is very important. There is no
requirement that a
> plugin use a predefined interface id. However, without
requiring a
> interface id, it causes several issues:
> 1) There's no way to create a mod of the plugin should
the original
> author stop developing the plugin. Sure you can create
a mod, but it
> will not play nice in the options (it will not disable
the original
> plugin when enabling the new one).
> 2) There's no way to disable a class of plugins in the
core without
> disabling it by plugin id if plugins don't provide a
valid interface id.
> While both methods of disabling plugins (by plugin id
and by interface
> id) will be available, there may be situations where we
change some apis
> and need to disable a whole class of plugins. If we
didn't require a
> interface id, this wouldn't be possible.
>
> So, the short of it is that a plugin should generate a
new interface id
> if there is no official one. We will use
newpluginapi.h to maintain a
> list of common interfaces (even ones outside of the
core).
>
> Rob
>
> Martin Afanasjew wrote:
>> Hi,
>>
>> if I understand correctly, the logic with SVN
commit 5041 is that the
>> MirandaPluginInfoEx function will only be used if
the plugin has a valid
>> MirandaPluginInterfaces function _and_ returns at
least one interface
>> not equal to MIID_LAST.
>>
>> I don't think that's reasonable because there are
plugins that offer
>> only end-user functionality but no interfaces that
can be used by other
>> plugins. They probably register services only to be
able to handle menu
>> items (and even this might be optional).
>>
>> So I think the module loading code should still
look at InfoEx even if
>> there are no interfaces. I know, this is still
optional, but we might
>> stumble later on and maybe fall.
>>
>> The particular line of code I'm talking about (in
newplugins.c) is:
>>
>> if ( pi &&
>>
((bpi->Info&&pi->cbSize==sizeof(PLUGININFO))||
(bpi->InfoEx&&pi->cbSize==sizeof(PLUGININFOEX)
&&validInterfaceList(bpi->Interfaces)))
>>
>> Maybe I'm only confused...
>>
>> Regards,
>> Martin (aka SloMo)
------------------------------------------------------------
-------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the
chance to share your
opinions on IT & business topics through brief
surveys-and earn cash
http://www.techsay.com/default.
php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Miranda-Develop mailing list
Miranda-Develop lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/miranda
-develop
|
