[LV2] Startup slloowwss dooooowwwwnn???

hermann meyer brummer- at web.de
Mon Oct 27 10:59:59 PDT 2014

Am 27.10.2014 14:56, schrieb Stefano D'Angelo:
> 2014-10-27 7:33 GMT+01:00 hermann meyer <brummer- at web.de>:
>> Am 27.10.2014 01:56, schrieb David Robillard:
>>> On Thu, 2014-10-02 at 17:19 +0200, hermann meyer wrote:
>>> I have been meaning to add a config file for lilv to be able to disable
>>> dyn-manifest (and thus naspro bridges) at run time.  Hosts can currently
>>> do this, but there's no easy mechanism to do it globally.
>>> Cheers,
>> Maybe a config switch were doubled plugs could be switched of generally,
>> instead print warnings about, would be good?
>> As far I remember right, the problem with naspro seems to be that it not
>> only add the ladspa plugs to the world, but scans again the existing lv2
>> plugs and double them up, which leads to a huge amount of error messages in
>> the terminal. My terminal cache was overflowed by the amount.
> FYI, NASPRO bridges scan LADSPA_PATH and DSSI_PATH (or default paths
> if not set). They won't scan LV2_PATH and won't "double" LV2 plugins.
> What you probably see are LADSPA plugins being loaded where you also
> have LV2 versions (e.g. swh, invada). In these cases there is a
> mechanism to avoid the "duplication" using dc:replaces in ttl files (I
> don't know if Lilv supports that automagically?). The bridges also
> don't print anything on stdout/stderr.
> BR,
> Stefano
As I said, this behavior I notice with a updated LV2 core and not 
updated naspro-bridges.
With updated naspro-bridges, I receive just a single error message.

> Warning: unknown parameter map_curve for plugin organ
> Warning: unknown parameter perc_timbre for plugin organ

But, the slow down on start-up a host application, eg. scan the world, 
is "horrible", even with updated naspro-bridges. And even more notable 
is, the slow down of a single plugin load. That is horrible, updated 
naspro-bridges, or not, with lv2-dev 1.10.0 and libliv-dev 0.20.0, 
naspro-bridges becomes unusable.


More information about the Devel mailing list