Add DOMException #1040
Add DOMException #1040
Conversation
| function bjson_test_fuzz() | ||
| { | ||
| var corpus = [ | ||
| "FBAAAAAABGA=", |
There was a problem hiding this comment.
Note: this was complaining because of the bytecode version, but I don't know how the corpus was generated so I just incremented it manually...
|
It's still missing a stack trace when created with the constructor... |
|
I was confused for a minute because browsers set |
|
Perhaps you could use Js_NewError and extract it? |
I think that should be fine. |
|
@bnoordhuis Any further comments on this one? |
Implemented separately from the other errors because it is defined in terms of WebIDL, where members of an interface are getters on their prototype. See the difference between `JSON.stringify(Object.getOwnPropertyDescriptors(new TypeError()))` vs `JSON.stringify(Object.getOwnPropertyDescriptors(new DOMException()))`. Note: the standard doesn't specify where to put "stack". We follow existing practice which imitates node instead of browsers.
|
@bnoordhuis Shall I land this? atob / btoa relies on it so maybe it's time for it :-) |
quickjs.c
Outdated
| JS_CGETSET_DEF("name", js_domexception_get_name, NULL ), | ||
| JS_CGETSET_DEF("message", js_domexception_get_message, NULL ), |
There was a problem hiding this comment.
These aren't wrong but if you want to DRY the code, you can do:
| JS_CGETSET_DEF("name", js_domexception_get_name, NULL ), | |
| JS_CGETSET_DEF("message", js_domexception_get_message, NULL ), | |
| JS_CGETSET_MAGIC_DEF("name", js_domexception_getfield, NULL, | |
| offsetof(JSDOMExceptionData, name)), | |
| JS_CGETSET_MAGIC_DEF("message", js_domexception_getfield, NULL, | |
| offsetof(JSDOMExceptionData, message)), |
See js_callsite_getfield for an example:
static JSValue js_callsite_getfield(JSContext *ctx, JSValueConst this_val, int argc, JSValueConst *argv, int magic)
{
JSCallSiteData *csd = JS_GetOpaque2(ctx, this_val, JS_CLASS_CALL_SITE);
if (!csd)
return JS_EXCEPTION;
JSValue *field = (void *)((char *)csd + magic);
return js_dup(*field);
}This is another one of those things we could generalize more because ultimately it just computes:
JSValue *field = (void *)((char *)JS_VALUE_GET_OBJ(this_val)->u.opaque + magic);There was a problem hiding this comment.
OK, done. (At this scale though, it seems a bit closer to golfing than DRY :P)
There was a problem hiding this comment.
Not even code golfing, I'm afraid... I tried adding a generalized getter to the whole code base and it added about as many lines of code as it shaved off. Not worth it at the moment, I guess.
| JS_EXTERN JSValue JS_PRINTF_FORMAT_ATTR(2, 3) JS_ThrowReferenceError(JSContext *ctx, JS_PRINTF_FORMAT const char *fmt, ...); | ||
| JS_EXTERN JSValue JS_PRINTF_FORMAT_ATTR(2, 3) JS_ThrowSyntaxError(JSContext *ctx, JS_PRINTF_FORMAT const char *fmt, ...); | ||
| JS_EXTERN JSValue JS_PRINTF_FORMAT_ATTR(2, 3) JS_ThrowTypeError(JSContext *ctx, JS_PRINTF_FORMAT const char *fmt, ...); | ||
| JS_EXTERN JSValue JS_PRINTF_FORMAT_ATTR(3, 4) JS_ThrowDOMException(JSContext *ctx, const char *name, JS_PRINTF_FORMAT const char *fmt, ...); |
There was a problem hiding this comment.
What does this function do if I don't call JS_AddIntrinsicDOMException first? Maybe it should have an "is initialized?" check?
(My guess is it throws a prototype-less object but I'm not 100% sure.)
There was a problem hiding this comment.
It's a user error. I've added an assertion.
|
Great work @bptato ! |
3 participants
Implemented separately from the other errors because it is defined in terms of WebIDL, where members of an interface are getters on their prototype.
See the difference between
JSON.stringify(Object.getOwnPropertyDescriptors(new TypeError()))vsJSON.stringify(Object.getOwnPropertyDescriptors(new DOMException())).(Required for btoa/atob and structuredClone; ref. #16, #1032)