| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-10-27 | 3.0 kB | |
| v1.2.0 source code.tar.gz | 2026-10-27 | 282.7 kB | |
| v1.2.0 source code.zip | 2026-10-27 | 310.2 kB | |
| Totals: 3 Items | 595.9 kB | 0 | |
CSV v1.2.0
Changes
selectanddropaccept an(i, name) -> Boolfunction again, as in 0.10 (#1210).
Fixes
- Delimiter detection only elects a delimiter that splits the first row, so a single-column file of times or timestamps is read as one column again (#1211).
- A parallel read with a
typemapwhose target is outside the promotion order (such asInt64 => Int32) could read a sparse column as text; it now matches a serial read (#1222). - A header with many copies of one name is no longer quadratic to read: 10,000 copies took 40 s and now take milliseconds (#1219).
Performance
- Text-heavy files read up to twice as fast with several tasks and use less memory (#1220).
- Wide files read 20-25% faster with several tasks (#1221).
- Quoted files use about 10% less CPU (#1223).
- Loading CSV no longer invalidates about 400 compiled Base method instances (#1224).
Documentation
?CSV.Filelists every reader keyword and?CSV.writethe writer keywords (#1215, [#1217]), and the migration guide's 0.10 comparisons are corrected (#1217).
(Release notes generated via AI.)
Merged pull requests:
- Document explicit types for numeric date formats (#1042) (@DylanModesitt)
- Buffer archive bytes in the ZIP example (#1166) (@nhz2)
- Document reading gzipped text in other encodings (#1209) (@quinnj)
- Accept (i, name) -> Bool functions for select and drop (#1210) (@quinnj)
- Preserve single-column values during delimiter detection (#1211) (@quinnj)
- List the reader keywords in the CSV.File docstring (#1215) (@quinnj)
- Show a period group mark in the decimal-comma example (#1216) (@quinnj)
- Tighten the docs and fix inaccurate claims (#1217) (@quinnj)
- Keep duplicate column naming linear (#1219) (@quinnj)
- Reserve each chunk's string bytes once (#1220) (@quinnj)
- Read each chunk's column types under one lock (#1221) (@quinnj)
- Re-parse a stale chunk under the column's current type (#1222) (@quinnj)
- Check a quoted field for inner quotes with one vector compare (#1223) (@quinnj)
- Drop File's eltype method, which invalidated Base code on load (#1224) (@quinnj)
- Don't fail CI when the Codecov upload fails (#1225) (@quinnj)
Closed issues:
- Allow user-provided row and field error callback functions for custom error handling (#522)
- Need a feature to handle an extra unit row (#936)
- Keyword argument consistency: missingstring -> missingstrings (#998)
- Dates not automatically parsed if they can be integers (#1040)
- CSV.Chunks possible memory leak (#1048)
- CSV.Chunks API (#1049)
- There is no clear method reading non-UTF8 gzipped file in example (#1091)
- "Missing" Values (#1107)
- CSV.Chunks splits file into uneven chunks (#1122)
Datetypes should not be inferred from column (#1131)- Non-US number formats:
delim = ','is not handled well (#1150) - help migrating to CSV v1.0 (#1207)
- BUG: No keyword arguments in docs for CSV.File or CSV.Read (#1214)