repl: support hidden history file on Windows #12207
Conversation
On Windows when REPL history file has the hidden attribute node will fail when trying to open it in 'w' mode. This changes the mode to 'r+'. The file is guaranteed to exists because of earlier open call with 'a+'.
| @@ -164,13 +164,22 @@ function setupHistory(repl, historyPath, oldHistoryPath, ready) { | |||
| } | |||
| } | |||
|
|
|||
| fs.open(historyPath, 'w', onhandle); | |||
| fs.open(historyPath, 'r+', onhandle); | |||
There was a problem hiding this comment.
Won't this break creating a new file if none exists? [1]
| } | ||
|
|
||
| function onhandle(err, hnd) { | ||
| if (err) { | ||
| return ready(err); | ||
| } | ||
| fs.ftruncate(hnd, 0, (err) => { |
There was a problem hiding this comment.
From what I am gathering, the mode change no longer cause overwrites (appends instead) and so we need to "flush" the file?
From the docs though, it sounds like it does writes and not appends, so this shouldn't be necessary?
'r+' - Open file for reading and writing. An exception occurs if the file does not exist.
There was a problem hiding this comment.
Previous implementation cleared the file once when opening it with w. This was the only place that would reset history file content - i.e. if one would set NODE_REPL_HISTORY_SIZE to value lower than 1000, then the history file would be trimmed when node starts. This ftruncate call here is to preserve this functionality.
|
That's a good idea |
|
Windows test fail unrelated. Landed in bb041ea |
|
Should this be backported to v6.x? |
|
Wherever this gets backported, we also need #12762. |
Landed this with #12762, LMK if that wasn't a good idea. |
|
LGTM |
On Windows when REPL history file has the hidden attribute node will fail when trying to open it in
wmode. This changes the mode tor+. The file is guaranteed to exists because of earlier open call witha+.Fixes: #5261
Checklist
make -j4 test(UNIX), orvcbuild test(Windows) passesAffected core subsystem(s)
repl