Bug #88012
closed
Merge / Compress CSS & Javascript
Added by Ralf Hübner over 5 years ago.
Updated almost 3 years ago.
Description
Here are described how the css/js merge and compress build the resulted filenames:
https://docs.typo3.org/typo3cms/SkinningReference/BackendCssApi/CssConcatenation/Index.html
I have more than one Server and they have all the same files(-sizes), same paths. However the resulted merge files at the servers are diffrent in filenames:
1. server:
merged-dbceb37e783c36e36e08a56eaa6f429d-0ac3333507d91db92390f719270f329f.css?1553494702
2. server:
merged-a0d765f8232ab798e19f30e4effe3c88-8916bf6b7d8a5e0191598561a3f26df7.css?1553494314
It seems the timestamp (filectime instead filemtime) of the files kicks in due to creating the hashs?
For example default.css:
The "merger" looks at "Geändert" instead of "Modifiziert".
If none of the concatenated/merged files are modified,
the merged-filename (hash) should stay the same?
stat default.css
Datei: „default.css“
...
Zugriff : 2019-03-26 14:41:19.000000000 +0100
Modifiziert: 2018-11-28 11:33:46.000000000 +0100
Geändert : 2019-03-26 14:41:09.000000000 +0100
Kind regards
Ralf
- Status changed from New to Needs Feedback
the merge/compress uses mtime, not ctime - and the filesize as well as the absolute filename: `$filenameAbsolute . $fileStatus['mtime'] . $fileStatus['size'];`
Are the absolute file paths the same on the servers, too?
Susanne Moog wrote:
the merge/compress uses mtime, not ctime - and the filesize as well as the absolute filename: `$filenameAbsolute . $fileStatus['mtime'] . $fileStatus['size'];`
Are the absolute file paths the same on the servers, too?
Hi Susanne,
yes, the paths are the same.
Can it be a side effect, maybe of bootstrap_package. The generation of css (theme-.css) from scss-files are on-the-fly?
- Status changed from Needs Feedback to Closed
Hmm. It seems core can't do much, here. When mtime is involved an deployment honors this (like for instance with an rsync), the filenames should end up the same on all servers. I think we should go ahead and close this issue for now, it is not clear how to proceed currently. In case this is still a problem, we may want to start with a fresh issue.
Also available in: Atom
PDF