Added a system to cache some data that can be used when recompiling the filters for the incremental data updates.
For now this only used by for the asset data source internal path compilation of the asset views.
In my test on a bad scenario this yielded a reduction 98% of the time it take to call UContentBrowserAssetDataSource::CreateAssetFilter when the cache is used.
Before the optimization the reference call took 26.3 ms
After the optimization the reference call took 0.3725 ms
Also this change fix an issue with my last change interacting badly with the alias data source by not using its customization of the package paths hierarchy.
#preflight 645a92f42d27fa25b3069003
#jira UE-183276
#rb Rex.Hill
[CL 25396753 by julien stjean in ue5-main branch]
* Plugins that only contain cooked content (and have no C++ modules) no longer show their "Content" suffix.
* Folders that only contain private cooked content are no longer visible in the asset or path views.
#preflight 642f2e3cb306e98c632f0195
#rb Jason.Stasik, Aditya.RaviChandran
[CL 24994922 by jamie dale in ue5-main branch]
#rb Julien.StJean, Matt.Peters
#jira UE-177133
#preflight 64072b19ba12ba64167e20e6
- In the last year we have started to make changes to the package format to allow data to be stored in the same file, but not actually inside of the package itself (See FPackageTrailer) and we might consider moving some other data blobs there in the future. This change has meant that there are code paths that can change things with the package file, that don't actually call save package, such as the process that virtualizes assets and can leave incorrect, stale data in the content browser tool tip.
-- This means when a user virtualizes an asset, the tooltip might indicate that it has not been virtualized and the size of the file on disk has not been changed until they browser to a different folder and browse back to refresh. This has caused people to believe that the virtualization operation has failed.
- This change hooks up UContentBrowserAssetDataSource to trigger a EContentBrowserItemUpdateType::Modified event when ever the package file changes on disk.
- During package save this could trigger multiple EContentBrowserItemUpdateType::Modified events for the same package (one from saving and one from the file being modified) but it looks like duplicate events already occur (if the packages are checked out as part of the save, this will introduce another EContentBrowserItemUpdateType::Modified event). There is a TODO in the code that suggests we might want to spend time removing duplicates events before broadcasting the delegate but when profiling I found that the overhead is quite low. In one test, saving 20 packages cost 4ms for the events triggered by the save and an additional 4ms for the events triggered by the file changing on disk. With the margins being so low it still doesn't look like a problem that is worth spending time on, however I did add a TRACE_CPUPROFILER_EVENT_SCOPE to the broadcasting of the delegate (UContentBrowserDataSubsystem::ItemDataUpdatedDelegate) so if it does become a problem in the future it will be easier to track down.
[CL 24556355 by paul chipchase in ue5-main branch]
Fixups for editor support classes for content browser code.
#jira UE-161932
#rb jason.stasik,jamie.dale
#preflight 631bd2a0d31788ea3ac03570
[CL 21935588 by robert millar in ue5-main branch]
Allow browsing to aliases if source asset is not visible
Allow aliases in collections
#rb dave.belanger,jamie.dale
#preflight 62d0b1ad8e4b379590da3397
#ROBOMERGE-AUTHOR: jason.stasik
#ROBOMERGE-SOURCE: CL 21115320 via CL 21115363 via CL 21115394 via CL 21115412 via CL 21115431
#ROBOMERGE-BOT: UE5 (Release-Engine-Staging -> Main) (v972-20964824)
[CL 21142427 by jason stasik in ue5-main branch]
#rb jamie.dale
#preflight 624f8a9a70773560e4f88654
#ROBOMERGE-AUTHOR: rex.hill
#ROBOMERGE-SOURCE: CL 19706140 via CL 19707819 via CL 19707829 via CL 19707846
#ROBOMERGE-BOT: UE5 (Release-Engine-Staging -> Main) (v938-19570697)
[CL 19708766 by rex hill in ue5-main branch]
#preflight skip
#rnx
#ROBOMERGE-AUTHOR: rex.hill
#ROBOMERGE-SOURCE: CL 19050950 via CL 19052025 via CL 19052037 via CL 19052053 via CL 19058939
#ROBOMERGE-BOT: UE5 (Release-Engine-Staging -> Main) (v918-19018356)
[CL 19066071 by rex hill in ue5-main branch]
Add callback for source data generation to asset views.
Add UObject enumeration to content browser API (By Rex Hill).
Intentionally reserve thumbnail pool sizes instead of flat 1024 reservation.
#jira UE-129795
#rb Rex.Hill
#preflight 615e0c51c5e9590001f4a22a
#preflight 615e4400f13022000126efda
#robomerge[BOT4] 5.0
#robomerge[BOT1] Main
[CL 17744889 by daren cheng in ue5-main branch]
#rb jamie.dale
#preflight 60c2b5d11ab0310001d12a7c
#ROBOMERGE-SOURCE: CL 16640821
#ROBOMERGE-BOT: (v831-16623017)
[CL 16640843 by rex hill in ue5-main branch]
Speculative fix for an issue of bogus data being passed into the callback, likely from the array being appended to mid-update
#jira UE-117392
#rb Francis.Hurteau
#lockdown Simon.Tourangeau
#ROBOMERGE-SOURCE: CL 16594535 in //UE5/Release-5.0-EarlyAccess/...
#ROBOMERGE-BOT: STARSHIP (Release-5.0-EarlyAccess -> Main) (v828-16531559)
[CL 16594560 by jamie dale in ue5-main branch]
This can cause a reentrancy issue where the item currently being renamed is removed via an update while we're still processing it
#rb Francis.Hurteau
[CL 16412079 by Jamie Dale in ue5-main branch]
