Re: [PATCH] Fixed some Crashes, a memory leak and some minor issues

[Andreas Winkelmann <ml awinkelmann de>]

On Thursday 12 June 2014 09:26:35 Andreas Winkelmann wrote:

Hi David.

> On Saturday 07 June 2014 10:50:59 David King wrote:

> > I have tidied up and pushed the first and second patches to a
> > wip/awinkelmann branch. I find it quite easy to trigger a warning when
> > toggling between browser modes (file list or album/artist list). If I
> > select a file in the file list, then switch mode and then switch back to
> > the file list, I can trigger the warning:
> >
> > ** (easytag:3333): CRITICAL **: Action_Select_Browser_Style: assertion
> > 'ETCore->ETFileDisplayedList != NULL' failed
> >
> > I do not have much time to look into the problem at the moment, but
> > hopefully a simple change will be enough to fix it.
>
> Oh, damn ;-)
>
> I think I have found the reason.
>
> Happens if there is more than one album in the Browser List and you switch
> to Album/Artist List.
>
> ----------------------------------------------------------------------------
> ----------------------- GList *ET_Displayed_File_List_By_Etfile (ET_File
> *ETFile)
> {
>     GList *etfilelist;
>
>     for (etfilelist = ET_Displayed_File_List_First (); etfilelist != NULL;
>          etfilelist = ET_Displayed_File_List_Next ())
>     {
>         if (ETFile == (ET_File *)etfilelist->data)
>             break;
>     }
>     ETCore->ETFileDisplayedList = etfilelist; // To "save" the position like
> in ET_File_List_Next... (not very good - FIX ME)
>     return etfilelist;
> }
> ----------------------------------------------------------------------------
> -----------------------
>
> If the for-loop does not find the ETFile - given via argument - after the
> for- loop etfilelist is NULL but set to ETCore->ETFileDisplayedList and
> trashes the already stored List. This seems to cause the trouble.
>
> At the moment I am not sure why even this value has to be "saved". And the
> Text "(not very good - FIX ME)" looks also strange ;-)
>
> A "if(etfilelist)..." fixes the bug.
>
> But I will try to get a better understanding what happens to get a final
> solution.

Ok, I think I got this fixed.

A patch is attached.

Once more the GTKTreeSelection "changed" signal caused some Trouble. It gets 
emitted for each row while gtk_list_store_clear() clears the BrowserList. I 
have stopped invoking Browser_List_Row_Selected() while this happens.

-- 
Andreas

Attachment: 0001-Fixes-problems-with-an-assertion-in-Action_Select_Br.patch
Description: Text Data