A novel editor should not be recommended for encoding safety merely because it can open many formats. The decisive test is what happens when a manuscript contains a character that the selected output encoding cannot represent. The editor should identify the affected location, protect the original from an unsafe save, and provide a clear path back after the author resolves the problem.
Build a controlled short-story sample
Make a duplicate of a roughly 1,500-character Japanese short-story sample. Insert one emoji in dialogue, one circled number in a list, one uncommon name character, and a long dash. Record the paragraph and surrounding words for each insertion. Keep an untouched UTF-8 copy as the reference.
The sample should be small enough to inspect line by line but realistic enough to include dialogue, paragraph breaks, and punctuation. Use exactly the same file for every candidate. Different samples make it impossible to distinguish product behavior from manuscript differences.
Select a restrictive target deliberately
Choose Shift_JIS as a test target only to trigger a meaningful representability check. The goal is not to force the final novel into Shift_JIS. It is to observe how each candidate behaves before a potentially lossy save.
Record whether the editor marks the precise characters, displays only a generic warning, silently substitutes another character, or writes the file anyway. A warning that gives no location transfers the search burden to the author. Silent substitution is a serious failure for this selection criterion.
Test protection before testing convenience
With unresolved characters present, determine whether automatic saving continues. Use only the duplicate sample, because a candidate may overwrite the file. Record the path and modification time before and after the warning.
The preferred behavior is understandable protection: the author can see which characters are at risk and the application does not silently replace them through an automatic save. Do not award extra credit for themes, outline panels, or export formats until this safety gate passes.
Test partial recovery, not only total deletion
Replace the emoji with an approved textual expression but leave the circled number unresolved. A useful editor should distinguish the resolved position from the remaining problem. Then fix the second item and verify that warnings clear in a way the author can understand.
Close and reopen the safely saved duplicate. Compare all four marked paragraphs with the untouched UTF-8 reference. Recovery is not proven by a warning disappearing; the text must survive a round trip with the approved wording and paragraph structure intact.
Score candidates on three observable outcomes
The first outcome is location: can the writer find every affected range without a manual search through the entire novel? The second is protection: does the editor prevent or clearly control an unsafe save while unresolved characters remain? The third is recovery: after correction, can the writer save, reopen, and see the intended text?
Record pass, fail, or not tested for each outcome. Do not turn an untested feature into a pass. If a candidate behaves differently with manual save and automatic save, record both paths separately.
Keep selection separate from conversion and diagnosis
This test chooses an editor. A complete Shift_JIS-to-UTF-8 conversion workflow requires its own source preservation, decoding check, save-as operation, and reopen comparison. Diagnosing already garbled text requires determining whether the failure occurred during decoding, saving, or display. Those are different jobs.
During selection, never damage the original manuscript to prove a warning exists. Work from a duplicate, and stop when the product would overwrite uncertain content. The ability to stop safely is part of the result.
Evaluate Rune Studio within published claims
Current Rune Studio materials describe reading several Japanese encodings, highlighting Unicode characters that cannot be represented in the currently selected encoding, notifying the user, and stopping automatic save while a problem remains. Those documented behaviors match the selection criteria above. They are not a completed test of this story sample and do not prove that every unusual character will behave identically in every file.
Run the controlled sample on the current version and preserve the result. If the behavior differs from the documentation, record the version, selected encoding, affected character, and whether the save was manual or automatic.
Recommend only after the reopened-file check
An editor earns a recommendation for this use case after it locates the hazards, protects the duplicate, and returns to a normal save state after correction. The final reopened file must match the intended story at every marked paragraph.
Begin with the emoji and circled-number sample, not a valuable master manuscript. Compare candidates with the same three outcomes and leave uncertain behavior marked as untested. See the current encoding-related scope on the Rune Studio product page.