does Stirling-PDF process asynchronously ?

[smehdux]

Hi,

We were looking to the sources and we've discovered that Stirling-PDF launches a unique unoconv process to convert files to PDF leveraging open office.

My question is: did anyone carry out a performance tests by converting dozens or hundreds of files ? How does it behave ?

I believe that the Stirling-pdf API doesn't support any kind of queue or a mechanism to deal with a huge number of requests because of - maybe - OpenOffice process which doesn't allow to launch multi instances of its process.

I look forward to read your reponses

[Frooodle]

We implement a max instance count for each operation depending on how heavy it is
We plan to dynamically change this but currently it's static

Due to the max instance count the job does 'queue' of sorts

[Frooodle]

https://github.com/Stirling-Tools/Stirling-PDF/blob/main/src%2Fmain%2Fjava%2Fstirling%2Fsoftware%2FSPDF%2Futils%2FProcessExecutor.java
L48

[smehdux]

https://github.com/Stirling-Tools/Stirling-PDF/blob/main/src%2Fmain%2Fjava%2Fstirling%2Fsoftware%2FSPDF%2Futils%2FProcessExecutor.java L48

I saw it is why I'm questioning your team, so if we build our version with let's say 3 instances, could the subsequent command (unoconv/LibreOffice) launches 3 Linux processes ? I'm not sure about launching too many LibreOffice instances.

[Frooodle]

If you set this count to 3 or for some reason run 3 spdf instances on same machine then yes it could run 3 libre office instances

[smehdux]

Thank you @Frooodle, I hope the next version will manage this limit in the config file.
should I create a feature request or suggest it ?

[Frooodle]

Please create a feature request ticket!

[smehdux]

I've created it => #2003