Clean up sdlmutex.inc #123
Conversation
Previously, these types were aliases to the un-typed pointer type, which meant that any pointer value could be assigned to them. They are now declared as distinct pointer types.
Best regards |
|
With this patch, |
Free-Pascal-meets-SDL-Website
left a comment
Free-Pascal-meets-SDL-Website
left a comment
There was a problem hiding this comment.
Please have a look at the change proposals.
Best regards
Matthias
| { The SDL mutex structure, defined in SDL_sysmutex.c } | ||
| PPSDL_Mutex = ^PSDL_Mutex; | ||
| PSDL_Mutex = Pointer; //todo! | ||
| PSDL_Mutex = Type Pointer; |
There was a problem hiding this comment.
This should better be
type
PPSDL_Mutex = ^PSDL_Mutex;
PSDL_Mutex = ^TSDL_Mutex;
TSDL_Mutex = record end;
right?
There was a problem hiding this comment.
I guess we can use the record notation for now and later have a separate PR where we get rid of all the faux TSDL_Something types and replace PSDL_Something with Type Pointer.
There was a problem hiding this comment.
I see,
you never stated your opinion on this topic (exact translation of opaque structs), which was discussed in issue #63.
In short: From my point of view, which admittetedly is heavily influenced by a translator-view, the translation of
/* The SDL mutex structure, defined in SDL_sysmutex.c */
struct SDL_mutex;
typedef struct SDL_mutex SDL_mutex;
is better reflected by
TSDL_mutex = record end;
rather than by
PSDL_mutex = type Pointer;
I understand, that formally it would not be necessary to declare the TSDL_mutex type if it is not used in the header code though, but wouldn'd call this necessarily faux'ed. I consider this kind of a dilemma.
What do you think?
There was a problem hiding this comment.
I didn't think much of it when the question was first asked, basically following a "if it works, don't break it" principle - but over time, I came to favour the type Pointer solution, since by avoiding having a TSDL_Something declaration in the first place, it's made quite clear the user is not expected to ever use the non-pointer type.
There was a problem hiding this comment.
I see your point, it would make use of the units safer in some regard. Well, then let's make both ways possible for now. An update of the code style guidelines has to be done then. Let me know, if I should do it.
We could think about a goal for a future version of the units to only favor the pointer-translation for consistency.
Best regards
|
Thanks for the clean up! I will merge this and update the version tag and the cheatsheet hint separately and directly to the master branch. This is quicker than pushing to this PR as these files are not included, yet. |
2 participants
This patch fixes the definitions for
PSDL_Mutex,PSDL_SemandPSDL_Cond, adds a proper definition forSDL_MUTEX_MAXWAIT, and restores original SDL doc-comments for the mutex functions.