Generate Project.toml with Pkg's custom key order, and update reference tests to v1.7

[tkf]

In Julia 1.7, Project.toml generated by PkgTemplates is not compatible with Pkg's sort order.

(@v1.7) pkg> activate --temp
  Activating new project at `/tmp/jl_ZsryYH`

julia> using PkgTemplates
 │ Package PkgTemplates not found, but a package named PkgTemplates is available from a registry.
 │ Install package?
 │   (jl_ZsryYH) pkg> add PkgTemplates
 └ (y/n) [y]:
...

(jl_ZsryYH) pkg> st
      Status `/tmp/jl_ZsryYH/Project.toml`
  [14b8a8f1] PkgTemplates v0.7.25

julia> Template(user = "tkf")("ExamplePackage")
[ Info: Running prehooks
...
[ Info: New package is at /home/arakaki/.julia/dev/ExamplePackage

shell> cd ~/.julia/dev/ExamplePackage
/home/arakaki/.julia/dev/ExamplePackage

shell> cat Project.toml
name = "ExamplePackage"
uuid = "db43d3c3-1afb-45d2-b722-e2cdd05cc17f"
authors = ["Takafumi Arakaki <[email protected]> and contributors"]
version = "0.1.0"

[extras]
Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

[compat]
julia = "1"

[targets]
test = ["Test"]

The order of the tables is different from how Pkg sorts them. If I update it with Pkg:

(jl_ZsryYH) pkg> activate .
  Activating project at `~/.julia/dev/ExamplePackage`

(ExamplePackage) pkg> add UUIDs
...

Then git diff shows me

diff --git a/Project.toml b/Project.toml
index 3d3b37e..922c5e8 100644
--- a/Project.toml
+++ b/Project.toml
@@ -3,11 +3,14 @@ uuid = "db43d3c3-1afb-45d2-b722-e2cdd05cc17f"
 authors = ["Takafumi Arakaki <[email protected]> and contributors"]
 version = "0.1.0"

-[extras]
-Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+[deps]
+UUIDs = "cf7118a7-6976-5b1a-9a39-7adc72f591a4"

 [compat]
 julia = "1"

+[extras]
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
 [targets]
 test = ["Test"]

This PR solves this issue with a custom write_project with the sort key taken form Pkg.

It also bumps the Julia version used for the reference test to check the bug this PR fixes. To make it work, I added a mocking for uuid4.

close #298

[codecov]

Codecov Report

Merging #335 (01a45bd) into master (cb02b34) will increase coverage by 0.07%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #335      +/-   ##
==========================================
+ Coverage   94.68%   94.76%   +0.07%     
==========================================
  Files          20       20              
  Lines         621      630       +9     
==========================================
+ Hits          588      597       +9     
  Misses         33       33              

Impacted Files	Coverage Δ
src/plugins/project_file.jl	94.11% <100.00%> (+4.11%)	⬆️
src/plugins/tests.jl	100.00% <100.00%> (ø)
src/plugins/ci.jl	97.46% <0.00%> (-1.27%)	⬇️
src/template.jl	98.42% <0.00%> (-0.02%)	⬇️
src/PkgTemplates.jl	100.00% <0.00%> (ø)
src/plugins/badges.jl	100.00% <0.00%> (ø)
src/plugins/git.jl	95.08% <0.00%> (+1.53%)	⬆️

---

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 cb02b34...01a45bd. Read the comment docs.

[nickrobinson251]

Thanks a lot!

A couple small suggestions that i'd appreciate.

It also bumps the Julia version used for the reference test to check the bug this PR fixes.

This will close #298, which has been on the wish-list a long while, so i'm very grateful for you doing it!

To make it work, I added a mocking for uuid4.

Just to check my understanding: this mocking is needed as the output of uuid4() is different in v1.6+ (post JuliaLang/julia#35872)?

[tkf]

Thanks for the review!

This will close #298

FYI I added close #298 to the OP to auto-close it. But feel free to remove it if you manage it differently.

Just to check my understanding: this mocking is needed as the output of uuid4() is different in v1.6+ (post JuliaLang/julia#35872)?

Oooh, this explains why. I was a bit puzzled why, but I agree that it explains it. (I initially speculated that the test may actually not be running because that 1.5 was not in the CI matrix before realizing the explicit include)

[nickrobinson251]

hmm, it seems deeleting the 1.5 job from the GHA workflow file might not be sufficient, as it seems the repo is somehow still expecting a 1.5 job to run and pass... I wonder if this is something that needs changing in the repo settings (to which i don't have access)

Perhaps @christopher-dG knows? And/or perhaps @oxinabox might be able to bump my permissions to allow me to access repo settiing to investigate?

[christopher-dG]

I've replaced the 1.5 required check in the repo settings with a 1.7 one :)

[nickrobinson251]

Thanks!