[PD-dev] How to free memory of t_symbol

Pierre Alexandre Tremblay tremblap at gmail.com
Thu Feb 22 15:38:02 CET 2024


This is by far the best explanation of symbols I’ve read, ever.

Thanks!

> On 22 Feb 2024, at 13:41, Christof Ressi <info at christofressi.com> wrote:
> 
> Hi,
> 
> Pd symbols are immutable and permanent. `gensym("foo")` looks if the symbol "foo" already exists; if yes, it just returns it, otherwise it creates a new symbol, adds it to the global symbol table and finally returns it.
> 
> You must never attempt to free a symbol!
> 
>> ```
>> t_symbol **x_weights_arrays;
>> t_symbol *x_biases_array;
>> ``` 
> Here you must only free `x_weights_array`, which is an array of `t_symbol*`, but not the `t_symbol*` elements themselves.
> 
> ---
> 
> Side note: since symbols are unique and persistent, they can be compared *by address*. In other words, two symbols are equal if they have the same address. This is different from ordinary C-strings which may reside at different memory locations and thus need to be compared with `strcmp()` (or equivalent functions).
> 
> Christof
> 
> On 22.02.2024 10:19, Alexandros Drymonitis wrote:
>> I have a data structure with a symbol and an array of symbols that store array names defined as:
>> 
>> ```
>> t_symbol **x_weights_arrays;
>> t_symbol *x_biases_array;
>> ```
>> 
>> When I'm done with them, I want to free the memory, but calling free(x_biases_array) doesn't seem to work, and once called, as soon as I try to do something else in Pd (like unlock the patch and choose an object), Pd crashes.
>> 
>> I have narrowed down the error to freeing these symbols, and I'm sure the crash is not caused by something else. I scanned m_pd.h to see if there is a function to free t_symbol memory, but didn't seem to find anything. This might be something obvious for C-savvy people, but I'm not one, so any advice is welcome.
>> 
>> 
>> 
>> 
>> _______________________________________________
>> Pd-dev mailing list
>> Pd-dev at lists.iem.at
>> https://lists.puredata.info/listinfo/pd-dev
> 
> 
> 
> _______________________________________________
> Pd-dev mailing list
> Pd-dev at lists.iem.at
> https://lists.puredata.info/listinfo/pd-dev






More information about the Pd-dev mailing list