Great improvements! Congratulations! On Tue, Aug 15, 2023 at 2:35 AM Artyom V. Poptsov <poptsov.art...@gmail.com> wrote:
> Hello Guilers, > > I'm pleased to announce Guile-DSV 0.7.0: > https://github.com/artyom-poptsov/guile-dsv/releases/tag/v0.7.0 > > This release replaces the old finite-state machine (FSM) implementation > with the new one that uses Guile State Machine Compiler (Guile-SMC) [1]. > The new FSM code is generated at build-time from PlantUML description > directly. > > See the full list of user-visible changes below. > > As the result of re-writting FSMs of Guile-DSV with PlantUML/Guile-SMC, > Guile-DSV gained some speed improvements. You can find an attached > LibreOffice Calc document with some benchmarking using around 5 million > lines of data (source: [2]) > > More detailed information about the data (lines/words/characters): > --8<---------------cut here---------------start------------->8--- > $ $ wc datasets/FeedGrains.csv > 498929 6010097 93054624 datasets/FeedGrains.csv > --8<---------------cut here---------------end--------------->8--- > > I took five "dsv" runs using RFC 4180 input format with "--summary" > option on Guile-DSV 0.6.0 and Guile-DSV 0.7.0 with the same options (and > disabled logging) and in some cases the new parser implementation is 15 > times faster that the old one. > > As a bonus Guile-DSV FSMs are now much nicer and easier to maintain. > The downside is that now Guile-DSV depends on Guile-SMC. > > > * What is Guile-DSV? > > Guile-DSV is a library that allows to parse Delimiter-Separated Values > format (DSV). It supports two flavors of DSV: Unix-style[3] and > RFC4180[4]. > > Also Guile-DSV is shipped with a program named "dsv" that allows to read > and process DSV format (including delimiter change and conversion from > one standard to another, formatting tables etc.) See README file for > usage examples. > > > * List of user visible changes > > Here's an excerpt from the NEWS file: > > --8<---------------cut here---------------start------------->8--- > * Changes in version 0.7.0 (2023-08-15) > ** Guile-DSV now uses Guile State Machine Compiler (Guile-SMC) > Guile-DSV now uses Guile State Machine Compiler (Guile-SMC) for the state > machine code generation from the PlantUML description: > https://github.com/artyom-poptsov/guile-smc > > ** =dsv->scm= and =dsv-string->scm= now accept additional parameters > Now =dsv->scm= and =dsv-string->scm= procedures accept additional > parameters: > - =#:debug-mode?= -- Enable/disable the debug mode. > - =#:log-driver= -- Set the logging driver. > - =#:log-opt= -- Set the logging options. > ** =dsv= tool now accepts =--log-driver= and =--log-opt= > ** Update Texinfo documentation > --8<---------------cut here---------------end--------------->8--- > > > Thanks and happy hacking! > > - Artyom > > 1. https://github.com/artyom-poptsov/guile-smc > 2. > https://www.ers.usda.gov/data-products/feed-grains-database/feed-grains-yearbook-tables/ > 3. http://www.catb.org/~esr/writings/taoup/html/ch05s02.html#id2901882 > 4. https://tools.ietf.org/html/rfc4180 > > > -- > Artyom "avp" Poptsov <poptsov.art...@gmail.com> > Home page: https://memory-heap.org/~avp/ > CADR Hackerspace co-founder: https://cadrspace.ru/ > GPG: D0C2 EAC1 3310 822D 98DE B57C E9C5 A2D9 0898 A02F >
