V2

[nicksnyder]

API ready for review.

To consider

• How to deal with backcompat of existing message files?
• How to version goi18n command?
• Change import path for v2
• Support vgo?

fixes #88
fixes #87
fixes #85
fixes #82
fixes #76
fixes #64
fixes #56
fixes #44
fixes #30
fixes #27
fixes #8

[codecov]

Codecov Report

Merging #92 into master will increase coverage by 3.2%.
The diff coverage is 66.37%.

@@            Coverage Diff            @@
##           master      #92     +/-   ##
=========================================
+ Coverage   59.92%   63.13%   +3.2%     
=========================================
  Files          14       27     +13     
  Lines        1138     2268   +1130     
=========================================
+ Hits          682     1432    +750     
- Misses        424      762    +338     
- Partials       32       74     +42

Impacted Files	Coverage Δ
v2/internal/message_template.go	0% <0%> (ø)
v2/internal/message.go	0% <0%> (ø)
v2/internal/parse.go	0% <0%> (ø)
v2/internal/plural/rule_gen.go	100% <100%> (ø)
v2/internal/plural/rule.go	100% <100%> (ø)
v2/goi18n/main.go	35.38% <35.38%> (ø)
v2/goi18n/extract_command.go	59.45% <59.45%> (ø)
v2/i18n/bundle.go	66.07% <66.07%> (ø)
v2/internal/plural/operands.go	67.64% <67.64%> (ø)
v2/goi18n/merge_command.go	74.43% <74.43%> (ø)
... and 16 more

---

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 e1e1f7b...a83f9c1. Read the comment docs.

[bep]

This looks really, really good. I assume that old bundle files will still work after this?

[nicksnyder]

I assume that old bundle files will still work after this?

No, I have not spent any time thinking about backcompat yet; I designed this fresh.

I think the best way to deal with old message files is to run a one-time conversion via goi18n. Thoughts?

[bep]

I think the best way to deal with old message files is to run a one-time conversion via goi18n. Thoughts?

I think there should be a really good reason to break the file formats.

Coming from Hugo, I mainly want this for the "non-globals". If I have to somehow handle migration of all the sites in the wild (I know I can probably point to some tool out there, but I still have to document it and answer questions from confused people), I may think twice before doing the upgrade.

[mh-cbon]

How to version goi18n command?

might this help golang/dep#221

[bep]

I think there should be a really good reason to break the file formats.

Note that I'm happy to do the work if it's a really good reason to, i.e. the change itself adds really useful stuff. But not just for vanity reasons ("this looks prettier").

[nicksnyder]

I think there should be a really good reason to break the file formats.
But not just for vanity reasons ("this looks prettier").

The file format (and everything else) in this PR was created from first principals, given the knowledge of file formats that we want to support, and feedback from users like this. There are also new things, like keeping track of the original content hash to know if a source translation was edited.

Supporting multiple file formats in general is confusing to users and a pain to maintain/test/document. The fact that v1 has multiple formats is an artifact of not understanding the limitations of alternative encodings (e.g. toml) from the beginning of the library.

How to deal with backwards compatibility is simply something that I have not thought about yet because backcompat was not a goal of mine when designing v2, but I will think about it soon. Of course I want to make the transition to v2 as easy as possible.

[nicksnyder]

@bep the runtime library can load message files from v1, and message files from v1 will automatically get reformated to the new structure the first time goi18n merge is run.