feat: support timestamp skipping in test cases

[r3v4s]

Description

support timestamp skipping in test cases

How has this been tested?

test case run by gnodev

[r3v4s]

This PR is ready, but I just saw #657
If GetTimestamp is going to be remove, we can close this PR

[thehowl]

@r3v4s I don't think we need to close it - can we make it use time.Now() instead?

[r3v4s]

@thehowl I wasn't thinking straight. Even if we get rid of std.GetTimeStamp we can still query timestamp using stdlibs's time package via time.Now()

So maybe we can leave this pr bit longer.

BTW, Can you please specify what you're referring to as "it" in "can we make it"?

[thehowl]

The PR. I was just saying you could modify the code to use time.Now instead :)

[moul]

To give you some context on time.Now(), currently it has a fixed value set during the first consensus and added to the transaction's metadata before execution. If you call it multiple times during the same call, you'll get the same value. However, we plan to make it dynamic by using opcodes for determinism. Regardless, it should be possible to replace std.GetTimestamp with time.Now() for the end developer.

[r3v4s]

NEW IDEA

Since #969 is merged, now gno will create block every 5 seconds.
How about adding ctx.Timestamp += (count * 5) in TestSkipHeights

gno/gnovm/tests/imports.go

Lines 643 to 656 in 5d3be42

	pn.DefineNative("TestSkipHeights",
	gno.Flds( // params
	"count", "int64",
	),
	gno.Flds( // results
	),
	func(m *gno.Machine) {
	arg0 := m.LastBlock().GetParams1().TV
	count := arg0.GetInt64()
	ctx := m.Context.(stdlibs.ExecContext)
	ctx.Height += count
	m.Context = ctx
	},

something like this maybe??

		pn.DefineNative("TestSkipHeights",
			gno.Flds( // params
				"count", "int64",
			),
			gno.Flds( // results
			),
			func(m *gno.Machine) {
				arg0 := m.LastBlock().GetParams1().TV
				count := arg0.GetInt64()

				ctx := m.Context.(stdlibs.ExecContext)
				ctx.Height += count

				ctx.Timestamp += (count * 5)
				m.Context = ctx
			},
		)

[thehowl]

Since #969 is merged, now gno will create block every 5 seconds.
How about adding ctx.Timestamp += (count * 5) in TestSkipHeights

I don't think block time every 5s is something set in stone, and I think having ctx.Timestamp += (count * 5) in TestSkipHeights is a bit "untransparent"? (even though it reflects real-world behaviour.)

Furthermore I think TestSetTimestamp (#924) might be a better feature.

[r3v4s]

@thehowl

• I don't think block time every 5s is something set in stone > Yep, it isn't set in stone
• "untransparent"? (even though it reflects real-world behaviour.) > Personally I just prefer for test case to be run as same as possible as real gno-vm. But I do understand this kind of automation can be opaque.

I'll still implement automation, but we can wait for #924

[netlify]

❌ Deploy Preview for gno-docs failed.

Name	Link
🔨 Latest commit	93f39c9
🔍 Latest deploy log	https://app.netlify.com/sites/gno-docs/deploys/65010fb8db87680008015060

[moul]

Looks good, but please add a unit test to prevent regressions.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 58.13%. Comparing base (dd68d61) to head (d4c8776).
Report is 104 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #569      +/-   ##
==========================================
+ Coverage   48.44%   58.13%   +9.69%     
==========================================
  Files         409      389      -20     
  Lines       61965    65275    +3310     
==========================================
+ Hits        30019    37950    +7931     
+ Misses      29446    24503    -4943     
- Partials     2500     2822     +322     

Flag	Coverage Δ
gnovm	60.00% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[leohhhn]

Hey @r3v4s - I see this is a long standing PR with some discussions going on on how we should do this.

To get a v1 on this - let's just simplify it and just add count to ctx.Timestamp. Then, please add a txtar test for this, and we can merge it. We should at least have something, and then iterate, rather than be stuck on a discussion.

Also, please check why examples/test is failing in the CI 🙏

[r3v4s]

@leohhhn

let's just simplify it and just add count to ctx.Timestamp

In this case, skipping 1 block will increase time 1 second. Did I understood correctly?

Then, please add a txtar test for this, and we can merge it.

I don't think I can. AFAIK Test* functions are only callable with gno test not txtar.

// UPDATE:
ci was failing because rand was using time value, fixed in cb8d0a0

[zivkovicmilos]

@moul I think @r3v4s added a regression test to the PR 🙏

[leohhhn]

Hey @r3v4s, I will try to push this forward with @moul. In the meantime, can you please add a sentence to this page to tell users that TestSkipHeight will also move the block timestamp?

[r3v4s]

Hey @r3v4s, I will try to push this forward with @moul. In the meantime, can you please add a sentence to this page to tell users that TestSkipHeight will also move the block timestamp?

Updated d4c8776

[zivkovicmilos]

Pinging @moul for a second review 🙏

Old, and issues addressed