Finish / fix price_to_string_test

[abitmore]

Code:

bitshares-core/tests/tests/app_util_tests.cpp

Line 240 in 25000ab

} catch(...) {}

Travis-CI reports:

3568856ms th_a app_util_tests.cpp:217 test_method ] i: 0 j: 0 p[i][j]: {"base":{"amount":-1,"asset_id":"1.3.0"},"quote":{"amount":-1,"asset_id":"1.3.0"}}

3568856ms th_a app_util_tests.cpp:239 test_method ] [ERROR: Got exception while trying to dump ( (price_to_string(p[i][j],0,0))(price_to_string(p[i][j],0,1))(price_to_string(p[i][j],0,2))(price_to_string(p[i][j],0,8))(price_to_string(p[i][j],0,19))(price_to_string(p[i][j],1,0))(price_to_string(p[i][j],1,15))(price_to_string(p[i][j],2,6))(price_to_string(p[i][j],2,10))(price_to_string(p[i][j],5,0))(price_to_string(p[i][j],9,1))(price_to_string(p[i][j],9,9))(price_to_string(p[i][j],9,19))(price_to_string(p[i][j],18,10))(price_to_string(p[i][j],18,13))(price_to_string(p[i][j],18,19))(price_to_string(p[i][j],19,0))(price_to_string(p[i][j],19,7))(price_to_string(p[i][j],19,19))(price_diff_percent_string(p[i][j],p[j][i])) )]

[jmjatlanta]

I realize this is low priority, but please include a hint as to what is left to be done. From my testing, the catch is never called, although the log shows some errors. Are you suggesting that we need to get rid of the idump and verify that an exception is thrown at the right time?

[jmjatlanta]

What should happen if the _price.base.amount is 0? price_to_string is currently returning the string "0", paying no attention to precision. But yet the next line asserts that _price.base.amount >= 0, which the = will never be true, as the return "0" above would have been hit. Should the FC_ASSERT be changed to _price.base.amount > 0? Or should the call ASSERT due to the _price.base.amount being 0?

bitshares-core/libraries/app/util.cpp

Lines 80 to 87 in fc2ab66

	string price_to_string( const price& _price, const uint8_t base_precision, const uint8_t quote_precision )
	{ try {
	if( _price.base.amount == 0 )
	return "0";
	FC_ASSERT( _price.base.amount >= 0 );
	FC_ASSERT( _price.quote.amount >= 0 );
	FC_ASSERT( base_precision <= 19 );
	FC_ASSERT( quote_precision <= 19 );

Also, should the rules be similar for price_diff_percent_string that is also being tested?

[pmconrad]

IMO the test needs special handling for i == 1 (result always equal to "0") and i == 0 || j == 0 (always throws).
The idump can be removed, unless you want to manually inspect the results in each test run. Better choose some reasonable sample for i and j and test these explicitly.

[jmjatlanta]

Fixed by PR #1518