Avoid usage of Rec.Count() if possible

[christophstuber]

Count can be one of the most expensive database operations on large tables. I often see examples like this in code:

Checking for no records:

if Rec.Count() = 0 then

Should be replaced by:

if Rec.IsEmpty() then

Checking for any records:

if Rec.Count() > 0 then

Becomes:

if not Rec.IsEmpty() then

Checking for exactly one record:

if Rec.Count() = 1 then

Refering to https://vjeko.com/2017/03/08/the-if-count-1-conundrum/

if Rec.Find('-') and (Rec.Next() = 0) then

Checking for more than one record:

if Rec.Count() > 1 then

if Rec.Find('-') and (Rec.Next() <> 0) then

[Arthurvdv]

@christophstuber, this idea was long overdue where I've today decided to create the rule for this.

I've ended up splitsing this into two separate rules. The reason for this, is that I believe the first one (LC0081) could be set as a warning for all projects, where the second one (LC0082) is a balance between performance and readability of the code. In this way it's more flexible on which of these rules want to be enforced.

• LC0081 - Use Rec.IsEmpty() for checking record existence.
• LC0082 - Use Rec.Find('-') with Rec.Next() for checking exactly one record.

Which both is now available in the v0.31.4 (pre)release version of the LinterCop.

Thank you for creating this idea, especially the part on the Rec.Find('-') with Rec.Next(), that is a nice catch.