Report a Typo

You do not need a local checkout, Python, or commit access to improve CDSL. A useful typo report gives maintainers enough evidence to find the exact entry, compare it with the printed scan, and decide which correction workflow applies.

What to collect

Start from the live site and write down:

Field	What to record	Example
Dictionary	The dictionary code or title	AP, MW, GRA
Headword	The word shown in the entry	arthaḥ
Location	Page/column, line number, or entry permalink if visible	printed page, scan page, or L number
Observed text	The text as CDSL currently displays it	kodhasyaiva
Expected text	The correction you propose	krodhasyaiva
Evidence	Why the correction is likely right	scan, parallel dictionary, grammar, citation
Transliteration	The script/scheme you used	Roman Unicode, Devanagari, SLP1

If the display has several modes, the Basic view is usually the easiest place to copy the headword and entry text. The S or scan link is the evidence trail back to the printed book; see Scans & Print.

Classify the problem

You do not need to choose the final label, but naming the likely class helps maintainers:

Looks like...	Usually means...	Example signal
Digital typo	The scan is correct, but the digital text has a typo	missing letter, wrong diacritic, broken word
Print correction	The printed book itself seems wrong	you can justify the correction from other reliable sources
Markup problem	The text is right, but display/linking is wrong	bad italics, broken citation link, visible tag
Scan problem	The digital text may be fine, but the scan is unusable	missing, blurry, skewed, wrong page
Scholarly question	The evidence is not decisive	conflicting sources or interpretation needed

Digital typos and markup problems can usually be corrected directly in the source text with an audit trail. Confirmed deviations from the printed book also need a printchange.txt record so future readers know the digital text intentionally differs from the scan.

Where to report it

For a dictionary-specific issue, open an issue in that dictionary's GitHub repository. The repository is usually named by the dictionary code or its project code; the catalog links each dictionary to its repository.

Include this checklist in the issue:

Dictionary:
Headword:
Observed text:
Expected text:
Where I saw it:
Scan / print evidence:
Other evidence:
Input/output transliteration:
Problem type if known:

If you have many corrections, group them by dictionary and source. Maintainers can then process them as a batch instead of as unrelated one-off reports.

What maintainers do next

Maintainers classify the issue using the Issue Taxonomy, verify the evidence, and then choose one of two paths:

• A single fix follows Process One Correction.
• A backlog or named contributor set follows Process a Scott Batch.

Both paths end in the same canonical reference workflow: source correction, XML validation, audit-trail change file, and paired commits to csl-orig and csl-corrections.