Rewrite, postpone and pause

[mlhetland]

This PR has three minor modifications to entr (cf. #298):

• By default, entr will now execute f() initially, unless the postpone keyword is used;
• entr sleeps for the amount given by pause between being triggered and executing f(), to handle clusters of modifications, such as text editor saves; and
• Rewrites to a given file/file name are now accepted as modifications; rewrites from the file name cause an error, as the file with the supplied name no longer exists.

The first and last are in line with the entr command line tool. The second is an added convenience, to avoid redundant execution when (e.g.) generating output based on saved files, etc.

The modifications are independent – no need to accept them all.

(Note: No new tests have been added for these features.)

[timholy]

I like all of these changes! Thanks for keeping them in separate PRs, when merged let's keep them that way.

Feel free to add any tests that you want. If some behaviors are difficult to test, I'll leave it up to your judgement to decide where to draw the line; let me know when you want this merged and I'll do it.

[codecov]

Codecov Report

Merging #308 into master will increase coverage by 0.02%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #308      +/-   ##
==========================================
+ Coverage   75.63%   75.66%   +0.02%     
==========================================
  Files          11       11              
  Lines        1133     1134       +1     
==========================================
+ Hits          857      858       +1     
  Misses        276      276

Impacted Files	Coverage Δ
src/Revise.jl	71.25% <100%> (+0.06%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3e14af8...69465c3. Read the comment docs.

[codecov]

Codecov Report

Merging #308 into master will increase coverage by 0.02%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #308      +/-   ##
==========================================
+ Coverage   75.63%   75.66%   +0.02%     
==========================================
  Files          11       11              
  Lines        1133     1134       +1     
==========================================
+ Hits          857      858       +1     
  Misses        276      276

Impacted Files	Coverage Δ
src/Revise.jl	71.25% <100%> (+0.06%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3e14af8...69465c3. Read the comment docs.

[codecov]

Codecov Report

Merging #308 into master will increase coverage by 0.02%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #308      +/-   ##
==========================================
+ Coverage   75.63%   75.66%   +0.02%     
==========================================
  Files          11       11              
  Lines        1133     1134       +1     
==========================================
+ Hits          857      858       +1     
  Misses        276      276

Impacted Files	Coverage Δ
src/Revise.jl	71.25% <100%> (+0.06%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3e14af8...69465c3. Read the comment docs.

[codecov]

Codecov Report

Merging #308 into master will increase coverage by 0.02%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #308      +/-   ##
==========================================
+ Coverage   75.63%   75.66%   +0.02%     
==========================================
  Files          11       11              
  Lines        1133     1134       +1     
==========================================
+ Hits          857      858       +1     
  Misses        276      276

Impacted Files	Coverage Δ
src/Revise.jl	71.25% <100%> (+0.06%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3e14af8...69465c3. Read the comment docs.

[mlhetland]

Feel free to add any tests that you want. If some behaviors are difficult to test, I'll leave it up to your judgement to decide where to draw the line; let me know when you want this merged and I'll do it.

There should probably be a couple of added tests in the "entr" test set. I've got it on my todo list, but if anyone is waiting for this fix/functionality, feel free to add/suggest some – not sure when I'll get to it.

[timholy]

If you think it will be a long time, like I said I'm willing to merge this now and risk breakage later. I kind of think of entr as "a community supported extra" to Revise, so I'm fine if the community is willing to risk breakage by going without thorough tests.

Just let me know what you want me to do.

[mlhetland]

I'm fine with that. I guess if there's a bug that needs fixing, we could add a test provoking that bug, anyway :-)

[timholy]

Technically this is breaking (previously we effectively had postpone=true), but I think of the old behavior as a bug, so I'll release this as Revise 2.2 rather than Revise 3.0.

[mlhetland]

Yes, I thought about that, too (specifically the postpone thing) – but I guess breaking changes don't depend on actual behavior, but documented behavior (if you're going with semantic versioning), so I guess the question here is how clear an expectation one might have, from the documentation, that there would be no call until the file changes. It states that it'll be called whenever the file is updated, of course, so I guess it would be reasonable to infer that it wouldn't be called until the file is updated post calling entr – but given the rather high-level description, it's not something I'd have relied on. (Then again, to find the specific behavior, I probably would have consulted the source, so … :-})

But despite being (mildly) breaking, the fix (postpone=true) is very easy to add. And, as you say, given the naming (from the command-line tool), it's not unreasonable to see this as a bug.

That reminds me: I didn't modify any documentation (beyond the docstring itself). If it's mentioned elsewhere, that should probably also be changed?

[timholy]

Improvement to the documentation would be most welcome. Note #311, which I think is due to JuliaInterpreter not working on nightly. (I guess fixing that should be my next priority.)