When .iml file is over 300k lines can't load project in windows.

Issue #624 resolved
Justin Julicher created an issue


I've just had to switch back to my windows 7 desktop on my work laptop and I'm now not able to load my projects if I refresh the metadata. Strangely I didn't have any issues with this running under ubuntu or suse linux.

Our org has upwards of 40k reports which causes all sorts of issues with metadata and we regularly have to exlude reports from our metadata.

Once I trim the file down I can then reload the reports.

Is there anyway to exclude certain types of metadata?



Comments (5)

  1. Scott Wells repo owner

    Justin, you can configure IntelliJ IDEA to handle this. First, I'd recommend giving it more memory as follows:


    and then for this specific problem, set the idea.max.content.load.filesize system property. The default is defined as:

    # Maximum file size (kilobytes) IDE is able to open.

    based on the file:


    You might bump that up to be larger than the .iml file that it's trying to load. What I don't know is how well it'll perform, but that should make it work for you.

    If that doesn't work, or if the performance is abysmal, let me know and I'll need to think about other options to help you get it working against that densely-populated org.

  2. Justin Julicher reporter

    Hi Scott,

    Thanks for the super quick reply! Yes that has fixed it. Performance seems to be ok and it's loading now.

    Although when figuring out the meta data to deploy it can take minutes or even 10's of minutes to get to the screen with the deployment details - I'm guessing if I remove the reports and the like that make the file so huge that it'll speed that process up?



  3. Scott Wells repo owner

    Well, it could be a couple of things causing those long running pauses. IC is using the metadata API to enumerate the metadata in the org to build the selector. It does cache that, but it also tries to be somewhat smart about when to read through that cache. It might be reading through when it doesn't need to. Either way I've long considered parallelizing the metadata queries for densely populated orgs like this. Maybe I'll throw that into an upcoming release and check with you on whether it makes much of a difference or not. I'll ping you when I do that.

  4. Log in to comment