rrlite

[richfitz]

• 1. What does this package do? (explain in 50 words or less)
     This package provides an interface to "rlite" - a standalone, zero-configuration port of the Redis interface (rlite is to Redis what sqlite is to MySQL).
• 2. Paste the full DESCRIPTION file inside a code block (bounded by ``` on either end).

Package: rrlite
Title: R bindings to rlite
Version: 0.1.0
Description: R bindings to rlite.  rlite is a "self-contained,
  serverless, zero-configuration, transactional redis-compatible
  database engine. rlite is to Redis what SQLite is to SQL.".
Depends: R (>= 3.1.2)
License: BSD_2_clause + file LICENSE
LazyData: true
Author: Rich FitzJohn
Maintainer: Rich FitzJohn <rich.fitzjohn@gmail.com>
Suggests: testthat,
    knitr,
    RcppRedis
Imports: R6, Rcpp
LinkingTo: Rcpp
VignetteBuilder: knitr
SystemRequirements: GNU make

• 3. URL for the package (the development repository, not a stylized html page)
     https://github.com/richfitz/rrlite
• 4. What data source(s) does it work with (if applicable)?
     User-supplied, but arbitrary. Special support for data.frames. No direct support for existing remote data sources.
• 5. Who is the target audience?
     Package developers, technically minded users who need to process large data or scale analyses. This is a low-level package.
• 6. Check the box next to each policy below, confirming that you agree. These are mandatory.
• The repository has continuous integration with Travis and/or another service
• The package contains a vignette
• The package contains a reasonably complete readme with devtools install instructions
• The package contains unit tests
• 7. Do you agree to follow the rOpenSci packaging guidelines? These aren't mandatory, but we strongly suggest you follow them. If you disagree with anything, please explain.
• Are there any packages that your package depends on that are not on CRAN?
• Do you intend for this package to go on CRAN?
• Does the package have a CRAN accepted license?
• Did devtools::check() produce any errors or warnings? If so paste them below.

* checking for executable files ... WARNING
Found the following executable files:
  src/rlite/deps/lua/src/lua
  src/rlite/deps/lua/src/luac

(this can be fixed by uncommenting the cleanup script but that slows down local development and I've not worked out a good workflow there. Rbuildignore has proved difficult to use for this).

• 8. Please add explanations below for any exceptions to the above:
     Not covered in the requested information, but in it's current state, the package is highly unlikely to work under Windows. I'm holding off attempting with R this until the upstream rlite is confirmed to work. Once that's done, this package would definitely want appveyor support to ensure continued Windows compatibility.

[sckott]

hey @jeroenooms

@richfitz package is our first to go through this process & acts as a test case to work out bugs before we start sending packages through the process. Do you mind reviewing this package? We'll check things that can be automated, including things that Ripley gets pissed about. Started a list of things for reviewers to check in the wiki https://github.com/ropensci/onboarding/wiki/For-Reviewers

@karthik @cboettig @richfitz @jennybc thoughts on the things for reviewers to check in the wiki?

[karthik]

@sckott The list is a good start. The last bit might be hard for a reviewer to judge since we haven't articulated that ourselves in policies. I've been taking a crack at that we should proceed with this one and answer questions as they arise. We can use that to keep iterating on that wiki.

[jeroen]

Sure, I'll have a look at it next week.

[gaborcsardi]

IMO a lot of these can be automated, so why don't we write a checkr package that does the automated parts, and then shows a nice summary. Parts of R CMD check can be used I guess, although its code quite involved and thus lengthy and somewhat unclear (to me).

[richfitz]

Ideally, because each package is on travis we can poll to get the most recent build logs? That gives all the R CMD check output without having to rerun anything. I agree that the more that is automated the better.

However, the items on the "for reviewers" checklist are subjective and not easily automatable. The intent here is that new packages to ropensci go beyond just working and are following and evolving set of best practices.

[karthik]

However, the items on the "for reviewers" checklist are subjective and not easily automatable. The intent here is that new packages to ropensci go beyond just working and are following and evolving set of best practices.

💯 to Rich's point. Nearly everything that needs to be automated already has. We need human oversight on fit and code quality.

[stewid]

@richfitz @sckott @karthik @cboettig @jennybc

Could the code be DRYed out? Is there a lot of code duplication?**

No

Are there user interface improvements that could be made?

No, the user interface was clearly described in the vignettes and intuitive to use.

Are there performance improvements that could be made?

No

Is documentation (installation instructions/vignettes/examples/demos) adequate?

The documentation is adequate with installation instructions that are easy to follow, two good vignettes and examples to all methods.

Does the package seem like a good fit for rOpenSci?

Yes 👍 It seems like a very efficient package to work with key-value data.

Other comments:

• Change title to title case in DESCRIPTION
• List author(s) of included software.
• List copyright(s) of included software.
• I was unable to run make check and make test. I'll open an issue with the output at https://github.com/richfitz/rrlite.
• Add argument check in e.g. https://github.com/richfitz/rrlite/blob/b390ba0afb5f1613572ba4ce3d5d733987edd0bd/src/rrlite.c#L35 to protect against unexpected values of filename (NA, NULL)

rrlite and git2r has chosen completly different strategies how to include and update the source code of the underlying interface. It would be interesting to have a follow up and list pros and cons with each approach for future guidance and best practice.

[richfitz]

Thanks, @stewid! 🌟

With packaging sources - I agree that there are pros and cons (there's a third approach in jqr that seems less robust than either). The git2r compilation also looks really different (https://github.com/ropensci/git2r/blob/master/src/Makevars.in)

Perhaps we could draft a blog post on "including 3rd party software in a package" and include best practices about listing authors, etc?

[stewid]

I'm using the Makefile https://github.com/ropensci/git2r/blob/master/Makefile#L62 to 1) copy files from a local libgit2 repository to git2r, 2) apply patches e.g. to pass R CMD check, and 3) create Makevars.in and Makevars.win from an R script https://github.com/ropensci/git2r/blob/master/scripts/build_Makevars.r

[karthik]

Perhaps we could draft a blog post on "including 3rd party software in a package" and include best practices about listing authors, etc?

💯 👏

[karthik]

@stewid If there isn't anything more for @richfitz to resolve, ok for us to move this into ropensci?
cc @sckott

[stewid]

Yes

[karthik]

ok great.
@richfitz Since you do have ownership of ropensci, go ahead and transfer it when you get a chance.
Other things to check off:

• Add a rOpenSci footer.
• Add an appropriate entry into ropensci.org/packages/index.html

[richfitz]

Sure - I was just going to squash the issues raised over here first.

One thing though; I was thinking about factoring out the auto-generated API (that'd be the "Intermediate" and "High" level interfaces in the rrlite readme) into a new package RedisAPI and leaving rrlite as a driver for rrlite. That would make the interface work more smoothly with RcppRedis and rredis. Would it be OK to transfer both halves into rOpenSci?

[karthik]

Right, missed that. Makes sense to do that first.

Would it be OK to transfer both halves into rOpenSci?

Sounds ok to me. @sckott any objections?

[jeroen]

Perhaps you can add a configure script with:

git submodule init
git submodule update

so dependencies will be automatically pulled in if not already? Do you plan to bundle a frozen version of the rlite source code when publishing to CRAN?

[richfitz]

Yes - I'll grab the current point of the rlite sources for CRAN. Probably once rlite stabilises I'll shift to pointing at its releases (right now it's a moving target).

[sckott]

no objections to putting both in ropensci, sounds good

[richfitz]

I've transferred ownership, though there a couple of remaining things in ropensci/rrlite#4

[cboettig]

👏 🚀

[sckott]

welcome on 🐗 d

[karthik]

@sckott you're 🔪ing me

[sckott]

:p