Mssql decimal fix

[hansnull]

Hi knex guys,

I have written a fix for large decimals in mssql, related to tediousjs/tedious#1058.
My fix adds dynamic scaling for decimal values and prevents an UInt64 overflow in tedious.

Best regards,
hansnull

[elhigu]

Please try to explain super clearly what is the problem you are now trying to solve here. What do you mean by dynamic scaling and why it is needed / wanted.

[hansnull]

I tried to store the following decimal value with knex and MSSQL: 10000000000005.74
No error was thrown, but MSSQL stores actually: -20037647.7924
The root cause for that is the following line:
https://github.com/tediousjs/tedious/blob/4120817754e91a32d9419093770771df590365ca/src/data-types/decimal.ts#L75
My fix calculates the scale and precision dynamically which fixes the problem for the number 10000000000005.74
To make sure that no other value overflows the UINT64 I added the check.

[elhigu]

Is this trying to recognize that if there is bigint coming from DB and it is too big to store in javascript number that error should not be thrown? 🤔

[hansnull]

Because javascript numbers are actually float values, the check is not necessary... I will remove it.

[elhigu]

https://github.com/tediousjs/tedious/blob/4120817754e91a32d9419093770771df590365ca/src/data-types/decimal.ts#L75

Cannot it be fixed in tedious if the root cause is there?

[hansnull]

It is also related to knex, because the default for the scale is 10 and for precision 38, which is not needed for a number like 10.01

[hansnull]

https://github.com/tediousjs/tedious/blob/4120817754e91a32d9419093770771df590365ca/src/data-types/decimal.ts#L75

Cannot it be fixed in tedious if the root cause is there?

I don't think so... see tediousjs/tedious#1058

[hansnull]

You can use my added unit test to compare the master version with my fix.

[hansnull]

All comments implemented

[elhigu]

Everything else seems to work just fine, except for typescript tests (npm run lint:types && npm run lint)seems to be failing for some reason.

[hansnull]

The failing tests seem not to be caused by my changes.

[kibertoad]

@hansnull Could you please pull latest changes from master?

[hansnull]

@hansnull Could you please pull latest changes from master?

Done

[kibertoad]

Released in 0.21.7