Re: [PATCH 3/4] vbscript: Implemented Sgn(try 2)
On 06/17/14 07:59, Shuai Meng wrote:
diff --git a/dlls/vbscript/global.c b/dlls/vbscript/global.c index c2b2edd..d21784d 100644 --- a/dlls/vbscript/global.c +++ b/dlls/vbscript/global.c @@ -1320,8 +1320,24 @@ static HRESULT Global_Int(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIA
static HRESULT Global_Sgn(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIANT *res) { - FIXME("\n"); - return E_NOTIMPL; + VARIANT v; + HRESULT hres; + + TRACE("(%s)\n", debugstr_variant(arg)); + + assert(args_cnt == 1); + + V_VT(&v) = VT_EMPTY; + hres = VariantChangeType(&v, arg, VARIANT_LOCALBOOL, VT_R8); + if (FAILED(hres)) + return S_OK;
Please use to_double() instead. to_double may need some fixes (it may require similar changes as you did for to_string first). Jacek
2014-06-17 17:43 GMT+08:00 Jacek Caban <jacek(a)codeweavers.com>:
On 06/17/14 07:59, Shuai Meng wrote:
diff --git a/dlls/vbscript/global.c b/dlls/vbscript/global.c index c2b2edd..d21784d 100644 --- a/dlls/vbscript/global.c +++ b/dlls/vbscript/global.c @@ -1320,8 +1320,24 @@ static HRESULT Global_Int(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIA
static HRESULT Global_Sgn(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIANT *res) { - FIXME("\n"); - return E_NOTIMPL; + VARIANT v; + HRESULT hres; + + TRACE("(%s)\n", debugstr_variant(arg)); + + assert(args_cnt == 1); + + V_VT(&v) = VT_EMPTY; + hres = VariantChangeType(&v, arg, VARIANT_LOCALBOOL, VT_R8); + if (FAILED(hres)) + return S_OK;
Please use to_double() instead. to_double may need some fixes (it may require similar changes as you did for to_string first).
Yes, to_double is better and simpler.
Jacek
2014-06-19 8:30 GMT+08:00 Shuai Meng <mengshuaicalendr(a)gmail.com>:
2014-06-17 17:43 GMT+08:00 Jacek Caban <jacek(a)codeweavers.com>:
On 06/17/14 07:59, Shuai Meng wrote:
diff --git a/dlls/vbscript/global.c b/dlls/vbscript/global.c index c2b2edd..d21784d 100644 --- a/dlls/vbscript/global.c +++ b/dlls/vbscript/global.c @@ -1320,8 +1320,24 @@ static HRESULT Global_Int(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIA
static HRESULT Global_Sgn(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIANT *res) { - FIXME("\n"); - return E_NOTIMPL; + VARIANT v; + HRESULT hres; + + TRACE("(%s)\n", debugstr_variant(arg)); + + assert(args_cnt == 1); + + V_VT(&v) = VT_EMPTY; + hres = VariantChangeType(&v, arg, VARIANT_LOCALBOOL, VT_R8); + if (FAILED(hres)) + return S_OK;
Please use to_double() instead. to_double may need some fixes (it may require similar changes as you did for to_string first).
Yes, to_double is better and simpler.
Jacek
I have tried to_double, but I don't think it is as convenient as
VariantChangeType. I have to add more types such as VT_CY and VT_EMPTY, and they don't work. For example, if 185 switch(V_VT(v)) { 186 case VT_EMPTY: 187 *ret = VT_EMPTY(v); but there is no VT_EMPTY(), how can I give ret whose type is double a "EMPTY" value? Why must we pick up to_double and give up VariantChangeType?
On 06/24/14 15:59, Shuai Meng wrote:
2014-06-19 8:30 GMT+08:00 Shuai Meng <mengshuaicalendr(a)gmail.com <mailto:mengshuaicalendr(a)gmail.com>>:
2014-06-17 17:43 GMT+08:00 Jacek Caban <jacek(a)codeweavers.com <mailto:jacek(a)codeweavers.com>>:
On 06/17/14 07:59, Shuai Meng wrote: > diff --git a/dlls/vbscript/global.c b/dlls/vbscript/global.c > index c2b2edd..d21784d 100644 > --- a/dlls/vbscript/global.c > +++ b/dlls/vbscript/global.c > @@ -1320,8 +1320,24 @@ static HRESULT Global_Int(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIA > > static HRESULT Global_Sgn(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIANT *res) > { > - FIXME("\n"); > - return E_NOTIMPL; > + VARIANT v; > + HRESULT hres; > + > + TRACE("(%s)\n", debugstr_variant(arg)); > + > + assert(args_cnt == 1); > + > + V_VT(&v) = VT_EMPTY; > + hres = VariantChangeType(&v, arg, VARIANT_LOCALBOOL, VT_R8); > + if (FAILED(hres)) > + return S_OK;
Please use to_double() instead. to_double may need some fixes (it may require similar changes as you did for to_string first).
Yes, to_double is better and simpler.
Jacek
I have tried to_double, but I don't think it is as convenient as VariantChangeType.
Not if you fix to_double first. Just call VariantChangeType inside to_double, like you did in to_string: http://source.winehq.org/git/wine.git/commitdiff/6bef4da592c8be4857dc0fa179a...
I have to add more types such as VT_CY and VT_EMPTY, and they don't work. For example, if 185 switch(V_VT(v)) { 186 case VT_EMPTY: 187 *ret = VT_EMPTY(v);
but there is no VT_EMPTY(), how can I give ret whose type is double a "EMPTY" value?
Well... of course there is no. As the name suggests, the type "empty" is empty and has no values ;) What will be result of converting empty to double is a matter of language semantic and vbscript happens to have the same semantic as VT_EMPTY->VT_R8 conversion in VariantChangeType.
Why must we pick up to_double and give up VariantChangeType?
I'm not saying about giving up VariantChangeType, I just want to use it right. Jacek
2014-06-25 0:50 GMT+08:00 Jacek Caban <jacek(a)codeweavers.com>:
On 06/24/14 15:59, Shuai Meng wrote:
2014-06-19 8:30 GMT+08:00 Shuai Meng <mengshuaicalendr(a)gmail.com>:
2014-06-17 17:43 GMT+08:00 Jacek Caban <jacek(a)codeweavers.com>:
On 06/17/14 07:59, Shuai Meng wrote:
diff --git a/dlls/vbscript/global.c b/dlls/vbscript/global.c index c2b2edd..d21784d 100644 --- a/dlls/vbscript/global.c +++ b/dlls/vbscript/global.c @@ -1320,8 +1320,24 @@ static HRESULT Global_Int(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIA
static HRESULT Global_Sgn(vbdisp_t *This, VARIANT *arg, unsigned args_cnt, VARIANT *res) { - FIXME("\n"); - return E_NOTIMPL; + VARIANT v; + HRESULT hres; + + TRACE("(%s)\n", debugstr_variant(arg)); + + assert(args_cnt == 1); + + V_VT(&v) = VT_EMPTY; + hres = VariantChangeType(&v, arg, VARIANT_LOCALBOOL, VT_R8); + if (FAILED(hres)) + return S_OK;
Please use to_double() instead. to_double may need some fixes (it may require similar changes as you did for to_string first).
Yes, to_double is better and simpler.
Jacek
I have tried to_double, but I don't think it is as convenient as
VariantChangeType.
Not if you fix to_double first. Just call VariantChangeType inside to_double, like you did in to_string:
Thank you, but when I use git blame, it turns out that it's you modify to_string, not me :)
http://source.winehq.org/git/wine.git/commitdiff/6bef4da592c8be4857dc0fa179a...
I have to add more types such as VT_CY and VT_EMPTY, and they don't work. For example, if 185 switch(V_VT(v)) { 186 case VT_EMPTY: 187 *ret = VT_EMPTY(v);
but there is no VT_EMPTY(), how can I give ret whose type is double a "EMPTY" value?
Well... of course there is no. As the name suggests, the type "empty" is empty and has no values ;) What will be result of converting empty to double is a matter of language semantic and vbscript happens to have the same semantic as VT_EMPTY->VT_R8 conversion in VariantChangeType.
Why must we pick up to_double and give up VariantChangeType?
I'm not saying about giving up VariantChangeType, I just want to use it right.
Jacek
participants (2)
-
Jacek Caban -
Shuai Meng