WebViewer Server 2019 Changelog

• Fix for a memory exhaustion bug that can cause crashes and out of heap errors
• Redesigned on-disk cache to deal with problems resulting from certain configuration options causing on-disk cache behaviour to not work as desired
• Improvements in CAD document handling
• Added ability to pass custom headers to GetPDF, GetThumb and PreloadURL.
• Fixes to container permission handling to allow usage with other Dockerfiles.
• Addition of TRN_DISABLE_VALIDATION option, disables certificate validation by WebViewer Server when fetching files
• Fix for multiple cached cookies of trnsockserv causing TRN_ENABLE_AUTHORIZATION to not work correctly
• Added appearance stream support for annotations
• Updated GetPDF.js to WVSApiWrapper.js, improved and changed structure of API
• Fix for files with annotations greater than 1 MB causing failures on document load
• Addition of TRN_AZURE_WEBAPP_COMPAT, this should be enabled if using Azure Web Apps, eliminates need for Azure Web App image variant
• Redesign of WVS building process, this has resulted in a reduction of image size from 6 GB to 3.7 GB
• Fix for CAD files not rendering properly on some server setups
• Fix for stickiness handling with HAProxy and other environments which use cookies for stickiness

• Memory Cache improvements to better respond to cache growth and adjustments
  • Now considers temporary directories when reading disk usage.
  • Improved disk usage metrics that trigger cleaning.
• Added cacheKey feature to allow caching of documents against a passed key.
• Update to PDFNet 7.0
• Fixed issue when preventing downloading of documents that were uploaded in a corrupt state to WebViewer Server
• Fixed error running container when no internal/static directories are defined.
• Added ability to pass a configuration file to WebViewer Server.
• Released the WebViewer Server Dockerless installer for Windows.
• Fixed critical timing issue that caused the first page of a document to return incorrect data.
• Fixed issue with TRN_FORWARD_CLIENT_COOKIES argument that caused it to improperly forward cookies as headers rather than cookie headers.
• All files are now linearized for to improve performance at a minimal cost. Previously, only files above 30 MB were linearized.
• Addition of the wvs_config.json configuration option to be used with the static installer.

• Changed build scheme within the WebViewer compose package. This package now pulls directly from DockerHub.
• Added TRN_FORCE_LOSSLESS_IMAGES for lossless image rendering to prevent artifacting when performing differential layering.
• Changed how SSL keys are placed into the server for HAProxy and Tomcat. They now use a volume mounting scheme and do not require a rebuild.
• Added ability to pass extra headers to GetPDF that will not be considered as part of the caching link.
• Several fixes to prevent errors that are caused by filenames being greater than 256 characters.
• Added fix for crash caused by .tif files.
• Corrected erroneous attempt to trigger a debug packager on server crash.
• Added default arguments for static and internal directories.
• Made all environment variables compatible with true and false and 1 and 0 to deal with inconsistencies.

• Fix for office documents sometimes causing fatal crashes when done in quick succession
• Added TRN_ENABLE_PER_SESSION_CACHING
• Revised TRN_ENABLE_SESSION_AUTH to be locked to a single session rather than any currently connecting one
• Minor fixes to environment config options

• Change in version number handling to match new versioning system
• Fix for crashes caused by timing error with cache cleanup
• Fix for filenames being too long and causing a render failure
• Increased the cacheMaxSize variable within Tomcat to deal with warnings resulting from large file caching
• Improved logging by removing non-informative outputs

• Fix for downloading files with encoded characters in their file names
• Fixed unicode names appearing incorrectly when downloading
• Changed Tomcat HTTP connector to APR from NIO

• Fix for URL Encoding so it more closely follows Chrome behaviour when fetching files
• Fix for in memory document cache memory leak

Version 1.2

New:

• Added support for CAD files in dgn, dwg, and dxf format.
• Can now limit backend document access to a given set of URL roots using the TRN_FETCH_REQUIRED_URL_ROOT option.
• Enabled automatic cookie forwarding for cases where the server and document store are in the same domain, using the TRN_FORWARD_CLIENT_COOKIES configuration option.
• Core logic now uses newest Apryse SDK 6.10 release.
• Server now supports a certificate whitelist -- useful for development environments using self-signed certificates.

Fixes:

• Improved support for very large files (>2GB) -- won't hand off rendering to the client in cases where it can't support it.
• Now properly support files with the .htm extension.
• Default load balancer cookie name now the same as HAProxy's cookie name without extra configuration.
• Various tweaks to how WebViewer client communicates with WebViewer Server, which should result in snappier performance for some client configurations.
• Fixed bug where we would incorrectly parse query parameters with no =.

• Fix for rendering html files

• Fix for missing backgrounds in HTML conversions
• Fix for overly encoded

links causing failed fetches

• Adjust CORS settings to properly support the XMLHttpRequest.withCredentials flag
• Updated SDK binary to address potential stability issues with very large log messages

• Document specific bug fixes