Changelog

What has changed in DaRT Reader, and what it means for the way you work with it.

DaRT Reader has a new graphical interface: the files of an extract, what it holds and its segments sit side by side, and after every write you see what was written and what the checks found.

New

• A new interface. Open a folder and its files are listed on the left. The extract you click is described in the middle: who created it, for which fiscal year, and whether its files are the size they should be. Its segments are listed on the right, where they can be searched, filtered by application, sorted and ticked for writing. Click a segment to see its fields, or its number of records to see its first 1,000 records.
• Writes run in a window of their own. It shows the progress of each segment and, at the end, what was written and what each check found. The results of database writes are kept under Run > Results.
• What the command line could do, the window can do too. Write to database can rename tables and fields and compare the extract's check sums against the written database.
• Help for support. Help > About DaRT Reader shows the version, the license status and where the settings and the log are, and copies them for a message to us.
• Settings that explain themselves. Each setting is explained beside it, and a value it cannot take is refused as you type.

Improved

• Large tables are read faster. A single large table is now shared among several readers instead of one. How many is set by Segments read at once: at 8, reading a table of 11,548 records drops from 35 ms to 8 ms. The setting also reaches as far in the window as on the command line; before, the window read at most five tables at a time.
• One log instead of eight files. The log window puts the logs of the window and of beye together in the order they were written, and narrows them by level, program and time.
• Renamed files no longer get lost. A payload file is recognised by its content rather than its name, so a renamed directory or payload file still belongs to its extract.
• More languages. Extracts in Japanese, Chinese, Korean, Thai, Vietnamese, Arabic, Hebrew and a number of other code pages can now be read by the installed program.
• Clearer findings. When the check sums of a group do not add up, the message names the table, the period, the company code and the field the group is formed by, with its value, and sets the database's sum and record count against the extract's.

Worth knowing before you update

• An extract opens only once the licensing service has confirmed your key, so opening one needs an internet connection. A demonstration key opens every extract and writes one record per segment; the window says so.
• Compare every value and Compare sampled rows have moved from the settings to the Write to database window.
• DaRT Reader 2.0 requires Windows 11, or macOS 26 on a Mac with Apple silicon.
• The installers now carry their version in the file name, such as dart-reader-2.0.0.msi, and the download page lists a SHA-256 checksum for each, so you can check that a download arrived complete.

Security and stability

A range of bug fixes (among them writing into an existing H2 database file, pausing a running write and writing to MS Access over several connections) along with libraries brought up to date.

This release is about two things: getting through an extract faster, and being able to show that what arrived in the database is what SAP wrote.

New

• Extracts are read faster. Several segments are now read at the same time instead of one after another, and the largest are read first, so a run is no longer left waiting on its biggest segment. How many at a time is a setting; the default suits most machines.
• Every segment is checked against the check sums the extract carries for it. A run succeeds only if the records read are the records SAP wrote — which catches what a record count cannot: a value quietly changed to another value of the same kind. A mismatch ends the run and names the segment.
• What was written is checked back against the extract. After a database write, two checks compare the result with the extract: one over every written value, one over a sample of written records, each naming where the two differ. Both are on unless you turn them off, and a difference is reported as a failed run rather than as a success — so an automated run can tell the two apart. Writing CSV files is unaffected.

Improved

• CSV files open correctly in Excel. The field separator and the decimal separator can now be chosen, and left unset they follow the region the machine is set to. The files are also marked as UTF-8, so umlauts arrive as umlauts on Windows.
• Amounts, quantities, rates and dates keep their form in the database. Decimal places survive as the extract declares them — three for quantities, five for exchange rates — and figures are written as numbers that can be summed and compared in SQL.

Worth knowing before you update

• The column types for amounts, quantities, rates and dates have changed, so a database written with this version does not have the same schema as one written with an earlier version. The types are listed in the manual.
• Table and field names are now written into SQL quoted. On a database that folds unquoted names — PostgreSQL to lower case, Oracle to upper — a table now arrives as TXW_FI_POS rather than txw_fi_pos, and a query saved against a database written by an earlier version may have to quote the name: SELECT * FROM "TXW_FI_POS". Databases that are already written are untouched.
• The check over sampled records adds two columns to every table it writes, DR_VOL_ID and DR_SRC_OFFSET, which are how a record is found again. They cannot be added afterwards, and anyone the database is handed to sees them. Turning that check off before writing means no table gets them.

Security and stability

A range of bug fixes and stability improvements throughout, along with updated database drivers.

This release collects everything since version 1.0.0, including the changes that were part of 1.0.1.

New

• Microsoft Access can now be used as a target database.
• A file chooser for file-based databases. For SQLite and Access you pick the database file in a dialog rather than typing its path.
• Choose how much is logged. The log level is a setting now and takes effect as soon as you save it, with no restart. The log itself opens from Help > View Logs.

Improved

• An incomplete write is now part of the result. If a segment could not be written to the database or to CSV, DaRT Reader names it when the run finishes, in the window and in beye alike. Previously the detail was in the log only.
• Settings and logs have moved out of the installation folder into your user profile. They survive an update, and writing them no longer needs administrator rights.
• German number and date formats are used again on systems set to German. Version 1.0.1 fell back to English in places.
• A licensing service that stumbles no longer ends the session. DaRT Reader retries before it gives up, and tells a rejected key apart from a service that is briefly unavailable, so you know whether to correct something or simply try again.
• The connection dialog and the settings dialog have been simplified, the extract views have a back button, and the field detail view has proper headings.
• Every log file now names the version that produced it, recorded at startup — so an empty log is a finding rather than a puzzle.

Windows

• Corrections to the installer: the version shown under Apps & features, the PATH entry, and the cleanup performed when you uninstall.

Security and stability

Updated database drivers, an updated embedded Java runtime, and a range of security and stability improvements throughout.