10

For example, look at this edit history:

https://meta.stackexchange.com/posts/359735/revisions

It looks like every single revision replaced all the text in the post, even when the revision comment says something like "deleted 1 character in body".

(I was unable to repro here so there is probably something more going on here.)

Examining revision 11 and revision 10 in my browser, they look fine; but fetching them with curl and doing a diff reveals that they are indeed both formatted as one loooooong line with the newlines represented as 
 (CR+LF). Why were the line breaks in that post saved as HTML entities in the rendered HTML, instead of line breaks? In the Markdown, they were typed as normal line breaks.

Improve this question
8
  • 1
    That first revision looks okay to me in side-by-side markdown, only inline and side-by-side seem broken. I indeed can't repro on the other post...
    – Tinkeringbell Mod
    Commented Jan 19, 2021 at 10:39
  • I partly repro'd: meta.stackexchange.com/posts/358670/revisions
    – Luuklag
    Commented Jan 19, 2021 at 10:43
  • @Luuklag That doesn't look like a repro to me; it only highlights the parts which actually changed.
    – tripleee
    Commented Jan 19, 2021 at 11:00
  • I tried some other posts containing tables and wasn't able to reproduce.
    – Sonic the Anonymous Hedgehog
    Commented Jan 19, 2021 at 11:10
  • I experimentally retyped all the newlines in the post in rev 12; let's see ...
    – tripleee
    Commented Jan 19, 2021 at 11:21
  • 1
    @Sonic Thanks for the edit; but do we know that Nicolas actually typed in that table? I would wager that he copy/pasted it from somewhere, possibly from Windows hell.
    – tripleee
    Commented Jan 19, 2021 at 11:30
  • The earlier diffs in meta.stackexchange.com/posts/359131/revisions until revision 39 look as if the whole table was replaced, but diffs for subsequent edits look just fine. Something in the diff algorithm must have changed on Jan 5.
    – Emil Jeřábek
    Commented Jan 19, 2021 at 13:16
  • 1
    Rev2 says "185 identical lines skipped" (expanded and SbS), which results from not diffing and counting correctly; where a huge difference is revealed by the inline view. I agree tripleee it's a virtual certainty that the table wasn't typed manually, it was generated and imported then exported - in a format that isn't strictly SE CommonMark, but it's good enough to display well, being processed by the Editor Diff is another matter entirely.
    – Rob
    Commented Jan 19, 2021 at 13:22

0

Reset to default

You must log in to answer this question.

Browse other questions tagged .