The LineDivider was made by @takashi, so I'm pinging him.

In general terms, you can try and work around memory issues by adding more memory, making sure the drive your FME temp is on has plenty of free space and, if you're not doing so already, run the 64-bit version of FME.

Another thing to look at would be to eliminate blocking transformers if possible, but it doesn't look like the LineDivider uses any.

@redgeographics is right. The LineDivider doesn't contain any blocking-transformer, it would never stores features into either memory or a file. I don't believe that it could cause a memory issue.

Firstly disconnect the LineDivider and the next transformer and run the workspace to see if the memory issue would still occur.

If the issue wouldn't occur, re-add the subsequent transformers one by one and run whenever adding, in order to determine where the memory issue occurs.

Otherwise, remove the transformers one by one from the LineDivider and run whenever removing, as well.

hi, @takashi @redgeographics I tried to add more space to store the temp feature but it didn't work. I am testing it by removing the transformers but it fails at line divider point only as it stores to feature in a temp file. In WB I using line divider to cut the lines based on length >> option as "shorten to the following length" and then I am going to use neighbor finder to compare the output from my line dividers. herewith I am attaching the screenshot to give an idea about the WB

Hi @takashi and @redgeographics

I am still facing difficulties with my workbench as it keeps crashing at line division point saying need more disk space. tried to run the WB on a machine with 64 GB ram and 1 TB of desk space.

A little background about the logic I use in this. I have two road layers from different sources and trying to find the missing road element in one of the input files. for this, I am calculating the angles dividing the lines using the line divider and converting it to point format . and then using the neighbor finder to find the matched and unmatched from base and candidates. then further I am separating this based on the angle. ANgle calculation step is done at the beginning of the wb.

Can somebody suggest a solution to this or maybe an alternate method to this analysis?

Thank you in advance.

How many features do you have in total in both datasets? Surely more than what's shown in your screenshot because that shouldn't faze a machine with those specs.

You can try tiling the data and process smaller sets.