Libs.adoc

Libs

My recommendations concerning various libraries.

• Google Guava: “new collection types (such as multimap and multiset), immutable collections, a graph library, functional types, an in-memory cache, and APIs/utilities for concurrency, I/O, hashing, primitives, reflection, string processing, and much more”. You should definitely use their preconditions helpers. Then check whatever else you might have a use for.

• Parse CSV files: FastCSV 2014 to 2024, very simple, 524 stars; commons-csv 2005 to 2024 and 374 stars; CSVeed 2013 to 2024 and using beans and 99 stars; flatpack 2005 to 2024 and 57 stars; SuperCSV is old; uniVocity is abandoned; more references

• Parse command line arguments: I have used Apache Commons CLI, but I’d encourage you to check out JCommander, or perhaps JOpt Simple.

• iCalendar (ical, ics) files: according to SO, biweekly is better than iCal4j. I found that iCal4j has very scarce documentation. A minor inconvenience of both is that they use old date-time java objects rather than the new java.time API.

• XML parsing: see DOM

• Graphs: Guava (see above) has nice Graph structures, but few algorithms.

  • JGraphT provides adapters to Guava and is very active.

  • JUNG inspired the structures of Guava; its most recent version is version 2.1.1, released 7 September 2016; “it’s active, it’s just proceeding very slowly”; this issue discusses its link to Guava.

  • GraphStream has a research team behind it, but I don’t like its core api: a node is a heavy interface that your own nodes have to extend. No apparent link to Guava.

  • Apache commons graph has gone unmaintained for a long time but was reactivated and committed to during June to Aug. 2020 then every few months from Oct 2022 (related discussions on the ML: still on sandbox ?, moving to git, Travis CI configuration, Build fails (unit tests))

  • Baeldung mentions “more powerful frameworks based on graphs, such as Apache Giraph, currently used at Facebook to analyze the graph formed by their users, and Apache TinkerPop, commonly used on top of graph databases”, which I didn’t check.

  • grph is not active any more

  • jgraphl is almost inactive

• Personaly, I refrain from using vʌvr (despite its very cool name), because it uses sneaky throws (this position is debated). Same for Faux Pas, NoException and fge’s throwing-lambdas.

• Native system information: OSHI

• JSON parsing: although I like spec-driven APIs, I do not recommend JSON-B anymore (I use Jackson but have not tried much alternative libraries). Here are the reasons.

  • The spec process is painfully slow. In June 2021 discussions started about supporting records officially; it was soon clear that implementations in fact did support them already, but more than three years later (Jan 2025), still nothing in the spec.

    • Jackson: similar discussion started in June 2021, issue completed in January 2023.

  • The spec wording is often unclear, it seems to not have been reviewed by fluent English speakers.

  • The discussions to improve the spec are painfully slow and non-converging. I have spent unreasonable efforts making a simple point (close() throws Exception, it shouldn’t), to no avail; here are other examples.

  • There is a huge difference between the number of users of JSON-B and its main alternatives (GSON and Jackson), judging by the number of artifacts in Maven central using them. This lets me think that JSON-B will evolve slower than the alternatives, so that the gap will grow.

  • More voices for Jackson VS JSON-B here.

  • In case you want to use it anyway, here is some (somewhat outdated) doc about JSON-B.