Fix fingerprint construction and comparison bugs

[juliusv]

No description provided.

[matttproud]

One caveat:

What about the stringified LENGTH in the format HASH-F-LENGTH-L? Will this handle said case appropriately?

[juliusv]

Question: String() is an alias for ToRowKey(), so even if "1000" < "2" by that definition, that ordering is still consistent with what's on disk, right?

[matttproud]

labelMatterLength -> labelMatterModulus?

labelMatterModulus := labelMatterLength % 10

Format string takes precision up to two digits? Then you can merely f.String() < o.String()?

[matttproud]

I believe that is correct. I would want to verify it, though.

2013/4/9 juliusv notifications@github.com

In model/fingerprinting.go:

@@ -151,20 +152,7 @@ func (f fingerprint) LastCharacterOfLastLabelValue() string {
}

func (f fingerprint) Less(o Fingerprint) bool {

• if f.Hash() < o.Hash() {

•   return true
  

• }
• if f.FirstCharacterOfFirstLabelName() < o.FirstCharacterOfFirstLabelName() {

•   return true
  

• }
• if f.LabelMatterLength() < o.LabelMatterLength() {

•   return true
  

• }
• if f.LastCharacterOfLastLabelValue() < o.LastCharacterOfLastLabelValue() {

•   return true
  

- }

• return false
• return f.String() < o.String()

Question: String() is an alias for ToRowKey(), so even if "1000" < "2" by
that definition, that ordering is still consistent with what's on disk,
right?

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/121/files#r3710121
.

[juliusv]

Well, the DTO fingerprint is just a proto version of String() / ToRowKey():

func (f fingerprint) ToDTO() *dto.Fingerprint {
  return &dto.Fingerprint{
    Signature: proto.String(f.ToRowKey()),
  }
}

and then the LevelDB key is built from that via:

targetKey = &dto.SampleKey{
  Fingerprint: fingerprint.ToDTO(),
  Timestamp: fooTime,
}

So my theory should hold. Ok to just keep my code as it is now then?

[matttproud]

Aside from that small nit, 👍