- Expose a previously silent download error (#110; Thanks to @jfy133!)
- Repair outdated links in documentation
- Improve some code sections in minor ways
All notable changes to this project will be documented in this file. The format is based on Keep a Changelog and this project adheres to Semantic Versioning.
- NEW API: The deprecations announced in v0.8.0 are now completed: The
bd_retrieve()
family of functions has replacedretrieve_data()
andretrieve_search_results()
has been renamed tobd_retrieve_by_search()
, see #96. Please update scripts and other downstream code that use BacDiveR. purrr
is no longer an imported / required dependency, but only suggested due to use in the introductory vignette, see #98.
- Warning message about no data being found is more informative now
- Documentation to follow BacDive's database updates
- Use pandoc 2.7.2 to generate the website
- NEW API: The
bd_retrieve()
family of functions will replaceretrieve_data()
soon. It consists ofbd_retrieve(id = …)
,bd_retrieve_by_culture(collectionno = …)
,bd_retrieve_by_sequence(accession = …)
andbd_retrieve_taxon(name = …)
which is more auto-complete-friendly and should help avoid the internal need to sanitise inputs.
retrieve_data()
has been replaced as described above.retrieve_search_results()
has been renamed tobd_retrieve_by_search()
.- Both old functions will be removed in the next major version. Please update your BacDiveR-using scripts and other downstream code.
- A bug in the test suite.
- Warnings, errors and message are now more concisely/informatively/strongly worded.
retrieve_data
&retrieve_search_results()
now validate all inputs (fix #71 & #78)
- The combined use of a) RCurl for downloading JSON-encoded data from BacDive with b) jsonlite for converting the download into an R objects was plagued by some whitespace escaping problems. This spawned a custom repair solution, adding accidental complexity. That swamp is now drained by use of httr.
- Instead of stopping with an error if a search returns no results, BacDiveR now only warns about this case and returns an empty list (#93).
- Estimated download times are now reported before all downloads longer than ca. 30 seconds (#84).
- Errors caused by comparing against a
NULL
(#91; Thanks to @jotech!)
- The vignette Logic-Checking BacDive Datasets
retrieve_search_results()
now returnsNULL
when no results are found, in order to ease integration of datasets intotestthat
tests.
- Another JSON whitespace escaping bug (#91; Thanks to @jotech!)
- Syncing of release versions to Zenodo, starting with 10.5281/zenodo.1308061
- The JSON downloads are no longer purged of all space characters pre-emptively
to prevent jsonlite from complaining about invalid encoding (#43). Instead,
only
\r
,\n
and\t
are repaired to\\r
,\\n
and\\t
, which jsonlite expects. This leads to different output (newline & tabs, where previously only spaces occurred)! Thus, if you are parsing BacDiveR output in any way, you may need to adjust that. Because I consider this unlikely given the "maturing" status, and because no API surface was changed, I don't consider this a major change in the SemVer.org sense.
- Install errors are now warnings and better advise about fixing them (#62)
- This changelog (see #41)
retrieve_search_results()
, see #61- Architecture Decision records, see #53 &
/docs/arch/adr-*.md
retrieve_data()
now downloads the dataset(s) by default, not only the ID(s), see #54 & #59- Datasets in the aggregated (large) list are now named according to their
bacdive_ID
s (see #47). Before, the lists were only numbered with1
,2
,3
, etc.
- An error in the download of a single dataset found through its culture collection number (see #45)
- Usable example / vignette in the README.md file (see #16)