Result limits and spill files

DataMind Oda keeps each step's result within a size a chat can carry; when more arrives than fits, it shows the part that stays, counts what it left out in a short note, and for the largest results stores the full text in a file on your computer. This page explains those notes, what each kind of result keeps, and where the rest goes.

Why a result gets shortened

• A result is a step's answer to Oda: the output of a command, the paths or matches a search found, the text of a page Oda read. When a result is larger than the chat can carry, Oda shows the part that stays and adds a note counting the rest.
• A note that counts entries in a list reads Omitted 3 items. A note that measures text reads Omitted 12 bytes., the size of the part left out.
• When the exact count is not known, the note reads More lines were omitted. and carries no number. When nothing was left out, there is no note at all.
• A note can join the count with the step's own advice on seeing more, for example: Omitted 25 items. Results capped at 100. Narrow the pattern, path, or include to see more.
• Two kinds of limit are at work: a count of entries for a list, and a size in bytes for text. A limit set to zero keeps none of the result and counts it all as omitted.
• A shortened result is not a failed result. When a note says text was truncated or omitted, it reports that a size limit was reached; it never means the source was incomplete or the step failed.

What each kind of result keeps

The part that stays depends on the kind of result:

• Cuts fall between characters, so a shortened result never ends on half a symbol.
• A read works differently: nothing is cut or stored. A long file comes back in parts, and each part says where to continue; Read, write and edit files covers the read windows.
• The searches' own counts and saved files are covered in Finding files and text; the notes a command result carries are in Shell commands.
• The web steps, with their result cards, are described in Web results in chat.

Where the full result goes

• The largest results are not lost to a cut: Oda writes the full formatted text to a file on your computer and shows a short preview with a note. The note names the file right after Full formatted result stored at:, and suggests how to work through it: Use read with offset/limit, or grep this path.
• The stored file holds the full result as it was formatted; for a fetched page it also keeps the line naming the page and its HTTP status.
• A search saves its complete list the same way: a file search writes glob-results.txt and a text search writes grep-results.txt, named in the result's note (Full sorted result stored at: for a file search, Full grep result stored at: for a text search).
• The note in the chat is the pointer to the file; the file stays on your computer, and Oda reads it back in parts or searches inside it when you ask. Where these files live, and when they are cleared up, is covered in Large results and spill files.
• Storing is best effort: when the file cannot be written, the result stays in the chat as it is and the step does not fail; a search result then says the complete result could not be saved and suggests a narrower search. A failed save never turns a good result into a lost one.

Questions, answered

Why does Oda cut long results instead of showing everything?

A chat carries a limited amount of text, so Oda keeps each step's result within a size it can carry. The part that did not fit is counted in a note, and where the full text can be kept, Oda stores it in a file and names it in the note. Nothing is silently lost, and a shortened result is not a failed one.

What does a note without a number mean?

The result was reduced to fit, but the exact size of the missing part is not known, so the note reads More lines were omitted. and carries no number. When nothing was left out, no note appears at all.

Why is only the end of my command's output shown?

That is how command results are kept: the end of the output stays, and a very large output keeps its start and its end as well. What was left out is counted in the note, and when the full output was saved, its note names the file. Shell commands lists the notes a command result carries.

Where does the rest of a long page go?

A fetched page keeps the start of the page in the chat; a very large one keeps its start and its end. The rest is not lost: Oda writes the full result, as it was formatted, to a file on your computer, and the note names it. Ask Oda to read that file, or to search inside it, when you need the rest.

Can a cut split a character in half?

No. Cuts fall between characters, so a shortened result never ends on half a symbol, and what you read is always made of whole characters.

Where does a search save its full list?

When a search finds more than its result keeps, the complete list is saved to a file: glob-results.txt for a file search and grep-results.txt for a text search, and the result's note names it. Finding files and text covers the counts each search keeps and its notes.

Why does a note tell me to narrow the pattern or path?

Some notes join the count with the step's own advice on seeing more. A capped file search, for example, suggests narrowing the pattern, path or file filter so a smaller, complete result can come back. The advice comes in the step's words, right after the count.

Is a long file read shortened like other results?

No. A read is not cut and stored: a long file comes back in parts, and each part says where to continue from. Read, write and edit files covers the read windows and the note that ends each part.