Path: intern new Strings

[jukzi]

When a filename is read from filesystem the time to intern() is not significant compared to the IO time but can save memory in the workspace

in case when i got an OOME it could have saved ~15MB

[github-actions]

Test Results

     24 files  ±0       24 suites  ±0   11m 27s ⏱️ ±0s
2 150 tests ±0  2 105 ✔️ ±0  45 💤 ±0  0 ❌ ±0 
2 194 runs  ±0  2 149 ✔️ ±0  45 💤 ±0  0 ❌ ±0 

Results for commit 0007c80. ± Comparison against base commit 9bb51e4.

[merks]

I imagine that here, like with EMF's URIs, massive duplication of strings are involved!

Not that it's a huge concern, but are interned strings actually garage collectable in modern VMs? (Sorry if that was answered elsewhere already.)

[jukzi]

are interned strings actually garage collectable in modern VMs

#33 (comment)

[jukzi]

There are only known TCK fails. Mind to reviev/submit any commiter, please?

[iloveeclipse]

Jörg, have you measured build time differences? We trade time vs memory here, and Path is heavily used during build.

[jukzi]

Jörg, have you measured build time differences? We trade time vs memory here, and Path is heavily used during build.

good that you asked: i checked and stumbled on a very ineffective org.eclipse.core.internal.watson.ElementTreeIterator.requestPath() which uses String concatenation just to split them afterwards again, which used intern():

[jukzi]

Let's drop this: it also interns ephemeral Paths
eclipse-platform/eclipse.platform@4e0f4a4 should be of much more help and we can investigate newer dumps for more potential - if needed at all.