Delayed updating of compiled content

[unwitting]

Me again, more lovely buggage. Steps to illustrate:

• src/test.jsx looks like:

const a = 'a';

• Run:

$ fusile src srv

(with or without --watch src/test.jsx)

• First read is as expected:

$ cat srv/test.js
'use strict';

var a = 'a';
//# sourceMappingURL=data:application/json;base64,eyJ2ZXJzaW9uIjozLCJzb3VyY2VzIjpbIi9Vc2Vycy9qYWNrcHJlc3Rvbi90bXAvc3JjL3Rlc3QuanN4Il0sIm5hbWVzIjpbXSwibWFwcGluZ3MiOiI7O0FBQUEsSUFBTSxDQUFDLEdBQUcsR0FBRyxDQUFDIiwiZmlsZSI6Ii9Vc2Vycy9qYWNrcHJlc3Rvbi90bXAvc3JjL3Rlc3QuanN4In0=

• Edit src/test.jsx to:

const a = 'abc';

• Next read is out of date:

$ cat srv/test.js
'use strict';

var a = 'a';
//# sourceMappingURL=data:application/json;base64,eyJ2ZXJzaW9uIjozLCJzb3VyY2VzIjpbIi9Vc2Vycy9qYWNrcHJlc3Rvbi90bXAvc3JjL3Rlc3QuanN4Il0sIm5hbWVzIjpbXSwibWFwcGluZ3MiOiI7O0FBQUEsSUFBTSxDQUFDLEdBQUcsR0FBRyxDQUFDIiwiZmlsZSI6Ii9Vc2Vycy9qYWNrcHJlc3Rvbi90bXAvc3JjL3Rlc3QuanN4In0=
=

• Next read is as expected:

$ cat srv/test.js
'use strict';

var a = 'abc';
//# sourceMappingURL=data:application/json;base64,eyJ2ZXJzaW9uIjozLCJzb3VyY2VzIjpbIi9Vc2Vycy9qYWNrcHJlc3Rvbi90bXAvc3JjL3Rlc3QuanN4Il0sIm5hbWVzIjpbXSwibWFwcGluZ3MiOiI7O0FBQUEsSUFBTSxDQUFDLEdBQUcsS0FBSyxDQUFDIiwiZmlsZSI6Ii9Vc2Vycy9qYWNrcHJlc3Rvbi90bXAvc3JjL3Rlc3QuanN4In0=

[Munter]

Was your update that didn't correctly get shown by any change made within 1000ms of the last time you updated the file? I have a known issue with the file system only providing second granularity on time, meaning I can't detect file updates within the same second. I probably need to adjust the caching mechanism to not trust cache within the same second as the current read

[unwitting]

Nah, can leave it as long as possible - first read is out of date, subsequent ones are fine.

[Munter]

Hmm :(

Do you have time to recreate the setup with that specific file in the fixtures and in https://github.com/Munter/fusile/blob/master/test/cache-test.js so there is a failing test? Pseudo code comments would be fine, just so I know what sequence I need to implement. I know the tests are not particularly to write with all the crazy race condition timeouts :/

[unwitting]

PR #24 - working test there, description copied below:

Submitted test - it actually now passes on my machine (as does my original manual case) but that feels very ropey so I think it's still best to pin it down with a test.