[LAD] [ANN] guitarix-0.04.4-1 release

hermann meyer brummer- at web.de
Sun May 24 13:56:03 UTC 2009


Am Sonntag, den 24.05.2009, 05:36 -0700 schrieb James Warden:
> Hi Hermann,
> 
> Great, I tested it against jack1 and jack2, and it is working fine now :)
> I looked at your additions. I have a couple of comments, nothing you should consider as important or critical:
> 
> - guitarix.cpp, line 529-530:
> I think any jack function will check if the client pointer passed as an argument is valid so it is safe to call jack_get_buffer_size before checking the client pointer validity. However, I have the habit of checking pointer validity before passing it to a 3rd party function. You never know, when the 3rd party lib is closed-source and its API badly documented,  you may get crashes. It's not the case here so you can keep it like that.
> 
> - guitarix.cpp, line 570-579: I would create a local function for this block of code because you call it in some other place as well. You will have only one place to maintain and gain more control on the variation of gNumOutChans. Turn it into:
> 
> if (merke == 0)
> {
>     gtx_unregister_ports();
> }
> 
> - german looking variable names: I have nothing against german, even learned it at school :D but it's open-source, not open-german-source  ;)
> 
> - I would also formalize a bit more what is guitarix's code so someone can tell right away which is your code and which is not. For example, start all your home made standalone functions with gtx_ (for GuiTariX)
> For home made classes, GtxSomeClass
> For class members of home made classes: fGtxSomeMember
> For class methods of home made classes: gtxSomeMethod()
> 
> This will become much more readable.
> 
> You don't have to though, and cleaning up is a bit of a tedious job ... but if it is to grow as you add more functionality, keep this in mind :)
> 
> Thanks for your work, I heard that you had to spend quite some time last night to make my patch fit in for jack1 ;)
> 
> Tschuess :)
> J.
> 
> 
Hello James

It have change again,the whole porthandling was a "dirty workaround" for
a proplem depper in the source. Fons give me some usefull hints to open
my mind and I be able to find and clear it. So the whole port
un/register stuff is gone, now guitarix only register ports when they
realy used (when jconv is run). A side effect from this, there is also
no different experience any more between jackd or jackdmp.

Okay, I will replace "merke" with "jcio" :D

   bye,bye
        hermann




More information about the Linux-audio-dev mailing list