Research Guides
Check Partial Downloads Before Citing a Complete Document
A download returns a successful HTTP status and a file appears in the research folder. Before calling the capture complete, check whether the response contains the whole representation or only a requested range. A partial transfer can be intentional and useful without being a complete research artifact.
This worksheet is for reviewing an authorized download workflow. It does not describe a Colquhoun.shop download feature or require changing the source server.
1. Keep transfer success separate from completeness
MDN describes HTTP 206 Partial Content as a successful response to a range request. For a single range, Content-Range identifies the returned portion. Multiple ranges use a multipart/byteranges response with range information for each part. MDN: 206 Partial Content
A review system should therefore retain both the observed status and a separate completeness decision. “Successful request” is not a sufficient label for “complete document ready to cite.” Equally, do not classify a legitimate range response as a broken source merely because it is not 200.
2. Read the range before naming the artifact
Imagine a fictional single-range response with Content-Range: bytes 0-999/5000. It describes 1,000 bytes from a 5,000-byte representation. Naming the saved body “complete-report” would overstate what that individual response supplied.
A browser or download client might obtain additional parts. Inspect the resulting artifact and the client's completion state rather than assuming one network row describes the final file. The purpose of the test is to establish what your research workflow actually retained.
Do not estimate pages read from the fraction of bytes received. Document structure and file encoding make that a different question. A 20 percent byte range does not establish that the first 20 percent of the readable report was reviewed.
3. Test with a known, non-private document
Use a small fixture you control, containing a clearly identifiable opening, middle and final section. Keep a known complete reference copy. Run the ordinary authorized download path and record whether the output is marked complete, partial or unresolved.
Open the retained artifact with the expected reader and inspect the final section as well as the first. A recognizable cover page is not enough. For structured research files, also check that the expected records and fields remain available.
If your workflow resumes interrupted downloads, test that path separately with the fixture. Do not manually concatenate arbitrary fragments and assume the result is valid. Retain the client's diagnostics for an implementer to review when the result cannot be established.
4. Copy this capture-completeness card
- Source: requested address and observation time.
- Request: ordinary download or range request, if known.
- Response: actual status and relevant range metadata.
- Artifact: saved file identifier and client completion state.
- Inspection: reader used and sections or records checked.
- Decision: complete for this task, partial or unresolved.
- Follow-up: authorized retry or technical investigation.
Keep private URLs and authentication details out of a shared card. Preserve diagnostic evidence in the appropriate restricted workspace rather than attaching an entire network export to a public report.
5. Carry the limit into the research handoff
If only an excerpt is available, label it as an excerpt and limit the conclusion accordingly. Do not write that a document contains no relevant passage when the unreviewed portion could contain it.
After obtaining a complete artifact, record that new result without erasing the earlier partial attempt. A colleague should be able to tell which file supports the final citation. That traceable handoff is the acceptance criterion, not merely a green request indicator.