Pricedb to stl #1731
Pricedb to stl #1731
Conversation
christopherlam
force-pushed
the
pricedb-to-stl
branch
7 times, most recently
from
b39e83f
to
599abfd
Compare
|
@christopherlam Re 2d730e3 and 6b741dd What's the point of having a PR--one marked Draft no less--when you're just going to push the commits directly to stable? BTW, I think the struct->std::pair change isn't a great idea on its own: It adds a bunch of overhead, and reduces expressiveness because "first" and "second" are less meaningful than "to" and "from". I didn't comment here because you have it marked draft and I was waiting for you to do something that would actually use some feature of the pair. |
|
Ok we can revert the |
christopherlam
force-pushed
the
pricedb-to-stl
branch
from
599abfd
to
99cee35
Compare
christopherlam
force-pushed
the
pricedb-to-stl
branch
2 times, most recently
from
e93c16a
to
5968e32
Compare
* use g_list_find_custom where appropriate * also g_list_free_full and g_list_foreach
christopherlam
force-pushed
the
pricedb-to-stl
branch
from
5968e32
to
fc21c55
Compare
| auto price_list = pricedb_get_prices_internal (db, c, currency, TRUE); | ||
| if (!price_list) return NULL; | ||
|
|
||
| for (auto item = price_list; item; item = item->next) | ||
| auto p = g_list_find_custom (price_list, GUINT_TO_POINTER(t), (GCompareFunc)price_time64_less_or_equal); |
There was a problem hiding this comment.
GUINT_TO_POINTER is an evil thing that subverts the type system. There are two problems in this case: First of all, time64s are signed and second they're int64_t. GUINT is an alias for uint32_t, and pointers on 32-bit builds (Windows!) are 32 bits. See the patch on bug 799104. The user doesn't explain themself well, but I guess they're doing forecasting and have dates past 2038 when time goes over 32 bits.
first easy steps.