opt(rdf-output): Make RDF output generation concurrent

[ahsanbarkati]

For RDF format output, the order of rdf's doesn't matter, so we can make them concurrent.

---

This change is 

[manishrjain]

Love it. The code is nice and simple. Yet, can be further simplified. Have some suggestions.

Reviewable status: 0 of 1 files reviewed, 5 unresolved discussions (waiting on @ahsanbarkati)

---

query/outputrdf.go, line 32 at r1 (raw file):

)

const numGo = 32

I'd just give it 4 goroutines.

---

query/outputrdf.go, line 43 at r1 (raw file):

// ToRDF converts the given subgraph list into rdf format.
func ToRDF(l *Latency, sgl []*SubGraph) ([]byte, error) {
	var fwg, wg sync.WaitGroup

what does fwg mean?

---

query/outputrdf.go, line 45 at r1 (raw file):

	var fwg, wg sync.WaitGroup
	b := &rdfBuilder{
		sgCh:  make(chan *SubGraph, 1000),

16 should be enough.

---

query/outputrdf.go, line 46 at r1 (raw file):

	b := &rdfBuilder{
		sgCh:  make(chan *SubGraph, 1000),
		outCh: make(chan *bytes.Buffer, 1000),

You could technically avoid this by just putting a mutex lock around the final buffer.

---

query/outputrdf.go, line 103 at r1 (raw file):

	defer wg.Done()
	for buf := range b.outCh {
		b.buf = append(b.buf, buf.Bytes()...)

This finalize isn't required. You could just create a write func in rdfBuilder, which takes a mutex lock and does the write from within.

That would also simplify how this code reads.

[ahsanbarkati]

Reviewable status: 0 of 1 files reviewed, 5 unresolved discussions (waiting on @ahsanbarkati and @manishrjain)

---

query/outputrdf.go, line 32 at r1 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

I'd just give it 4 goroutines.

Done.

---

query/outputrdf.go, line 45 at r1 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

16 should be enough.

Done.

---

query/outputrdf.go, line 46 at r1 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

You could technically avoid this by just putting a mutex lock around the final buffer.

Done.

---

query/outputrdf.go, line 103 at r1 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

This finalize isn't required. You could just create a write func in rdfBuilder, which takes a mutex lock and does the write from within.

That would also simplify how this code reads.

Done.

[manishrjain]

Feel free to merge.

Reviewed 1 of 1 files at r2, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @ahsanbarkati)

---

query/outputrdf.go, line 148 at r2 (raw file):

func (b *rdfBuilder) write(buf *bytes.Buffer) {
	b.Lock()
	defer b.Unlock()

you could just move unlock to after b.buf. Avoid using a defer. Micro-opt.