Re: A92: FileLIB functions


[Prev][Next][Index][Thread]

Re: A92: FileLIB functions




Well, I'm not all that of an experienced programmer, but I think you
should use the first way, but I don't know about how you would want to
pass the values to it.  I sort of lean towards using the stack. 
Experienced ASM programmers (not myself) would probably have now problem
doing it, and the rest of us would learn.  It just seems more flexible to
me for some reason.

     Daniel Plaisted
     -dsplaisted@bigfoot.com

On Fri, 9 Jan 1998 00:54:16 +0100 Arno.Kizina@t-online.de (Arno Kizina)
writes:
>
>Famille SCHERRER schrieb:
>>
>> 	Hello !
>>
>> I will make a library with a files & folders operations
>>  (create/delete/move/copy...)
>>
>> But I want to know WHICH posibility you prefer for the parameters...
>> There is two possibility to pass them :
>>
>> * With the stack.
>> 	So there is not destroyed registers (except the return value).
>> 	But it can take more memory, and it is not always the more
simple way.
>>
>> * With the registers.
>>
>> What do you prefer ???
>I prefer the use of registers, as dealing with the stack always includes

>thinking of the amount of parameters and their size, and it's faster.
Registers 
>to save must be saved by the user, so he has got the task to carefully
think 
>about usage.
>> And for the parameters, I would also know your opinion about WHICH
parameters
>>  would be the best :
>>
>> An example : the rename function.
>> three possibilities :
>> *	Index of folder where is the file
>> 	Index of file to rename
>> 	New file name
>>
>> *	Name of folder
>> 	Name of file
>> 	New file name
>>
>> *	Index of folder
>> 	Name of file
>> 	New file name
>>
> I think that the first is better. (there will be a find function, which
>>  return the index of the file/function defined by its name)
>> It is the easier way for browsers, isn't it ?? And after with the find
>>  function you will be able to make the other possibilities.
>I think the second is better because giving a pointer to the name is
easier 
>than first giving the pointer to the name and then providing and storing
the 
>handle(s).
>> COULD YOU ANSWER ME PLEASE ??? 
>> My e-mail address is p.c.scherrer@wanadoo.fr
>>
>> THANKS !
>> Benoit SCHERRER
>P.S. Do you really think that writing routines like yours is actually 
>necessary 
>in view of and waiting for ti92+, as ti MUST provide all of these 
>routines with 
>the ti92+?? I myself would keep my compressing routines in mind and 
>perhaps 
>write them as subroutines for general usage without regard to fargo, 
>which will 
>soon die out.		A.K.
>
>


References: