Documentation Center

Fixed Issues in 11.0.0 Web/App/Db Server

The following issues are fixed in version 11.0.0 of SDL LiveContent Architect compared to 10.0.0.

API

  • DocumentObj25.RetrieveMetadata fails with "An unexpected system failure occurred. The combination of field, level and operator must be unique". We now allow up to 9999 filters on the same combination. [TS-4850]
  • InfoShareWS AppPool crashes with a An unhandled exception occurred and the process was terminated. because of a Object reference not set to an instance of an object. in BaseErrorHandler.LogException. This issue was about adapting the code to no longer crash the AppPool, and potentially log the originating out-of-bound exception. [TS-8046|SRQ-1395]
  • Baseline25.wcf.CompleteReportByMode cannot auto-complete an image having multiple languages. This was caused by ISHBaselineReport_Out being to strict and not allowing the reportLanguage to be multi value as required by images relying on language applicability. [TS-7790|TS-7167]
  • Developing, the exception classes from API20 and API25 over WCF were missing user info like <logger applicationname="InfoShareAuthorQA02Sql2008R2" userid="GLOBAL\otokar" serverrequestid="645" serverhostname="MECDVTRI10QA02">. [TS-7326|TS-7325]
  • API25 like operations on LovValues in for example a DocumentObj25.Find function using the element name crashes on Oracle database based systems with the message: ORA-00903: invalid table name. [TS-6326]
  • Checkout fails for documents when SDL LiveContent Architect checks the value of the @data attribute. There is an issue when the value contains a character sequence which can be recognized as a number. If this condition is encountered, the characters are interpreted as an exponential number. For example, @data contains the letter e preceded by numbers, in this case the exponential value equals 6e93143 which results in a ORA-01426: numeric overflow. This happened on Oracle system and the com.p package file - submitted by DBUpgradeTool - has been tuned to no longer recognize character sequence like that as a number. [TS-6460]
  • Creating 100 or more card reference on Oracle, or 500 or more on SQLServer resulted in an Object reference not set to an instance of an object. exception. The most plausible scenario for this to happen is creating 100 sub folders within one parent folder on Oracle. The fix is a correction in the data layer that batches database inserts. [TS-6793]

Publishing

  • Whitespace missing when conditionalized <ph> phrases are filtered during publishing. [TS-8544|TS-8577|SRQ-2049]
  • Publishing PostProcessing IshPublUtil.vbs will no longer log statements Removing fixed attributes from file :... and setting ProhibitDTD to False which will reduce log size. [TS-6828]
  • Synchronizing and publishing to SDL LiveContent Reach will do parallel uploads. [TS-6586]
  • When publishing a sizeable publication to SDL LiveContent Reach, you might see that the PowerShell engine hangs without any progress. This is because the logging buffer of PowerShell requires a read at least every 4KB of data. This is a correction on a change introduced in 10.0.4. [TS-8195|TS-8200]
  • When publishing to SDL LiveContent Reach, the EventMonitor logging was missing the PowerShell engine log pieces. [TS-8047|TS-8086]
  • When publishing to SDL LiveContent Reach with extra resources like *.ZIP and *.PDF files, the memory consumption of the PowerShell engine could rapidly increase. By implementing more ByRef parameters the memory foot print was dramatically reduced. [TS-8016|SRQ-1357]
  • Publishing PostProcessing file IshPublUtil.vbs function ResolveHyperlinkFileNames doesn't reinitialize itself properly. [TS-8003|SRQ-1285]
  • Output format IshWebHelp has received the necessary Carbon 2.0 layout changes. [TS-7592|TS-7591]
  • When publishing to the "DITA XML" output format, the proprietary @cid attribute, used by the commenting engines, was not removed. This resulted in invalid xml files when verified with the standard OASIS DTDs. [TS-6570]
  • Publishing ended unexpectedly when empty hrefs were encountered. [TS-3367]
  • Support internal cross references in combination with DITA's copy-to attribute. One of the steps in preparing the DITA content for DITA-OT is replacing the GUIDS with filenames, this corrupted the internal defined cross-references in files which had been copied to other filenames. [TS-4849]
  • When something goes wrong during the publishing process (XML data error, Style sheet error, ...) and publishing fails, the same process may be restarted which results in a looping publication job. The job keeps running, consuming resources and affecting other jobs. [TS-5251]
  • Extended ServiceReferences UpdateContent, UpdateMetadata and UpdateProductDefinition with filename overloads to allow ChangeTracking Compare when publishing to SDL LiveContent Reach. This allows customization to step away from the previously hardcoded filename generation in the Trisoft.Utilities.ServiceReferences assembly functions. [TS-6073]
  • The transtype ishhtmlhelp was added to support non-Western languages CHM generation (e.g. Japanese). To generate non-Western chm files, configure the DITA-OT transformation type of the CHM (Compiled Help) output format to use ishhtmlhelp instead of htmlhelp.
  • The field, FISHKEEPFIXEDATTRIBUTES, was added to improve publishing performance. In the user interface this is seen as a new checkbox, Keep default attributes, on the output format. This option, when checked, means that the attributes which have fixed defaults in DTD are added to the XML file when they are exported during publishing. This results in the DTD not needing to be loaded again, since the needed attributes are in the XML document at publish time.

Translation Builder and Organizer

  • Translation Organizer service table ISH_TRANSJOBITEM has columns ISHTYPE and EDT that are inconsistent in length and collation. [TS-7389|TS-7367|TS-7388|TS-7300]
  • TranslationOrganizer diagnostics improved for template retrieval allowing to trace every retrieved template, every template with/without a mapped source language, every target language (not) mapped. [TS-7222]
  • TranslationOrganizer used to print out the locale information and not the language information. This is changed now and you will see the language information when triggering StartCheckWorldServer.bat. [TS-7223]
  • Documentation received an update on the flexibility of status transitions for TranslationOrganiser. [TS-7112]
  • Translation Jobs holding a publication skipped all children of an item if that item is not available in the requested language(s). Now any target language topics are added to the translation job even if for example the parent map still isn't released in its source language. [TS-6652]
  • Translation Job Properties drop down for Worldserver Translation Templates appears to be empty, but really isn't. Box is empty but if you go through you see different values but no text. Initially there should only be one empty item in the templates drop down (if no source language selected). When selecting a source language for which there are templates, they should appear in the templates drop down. If there are no templates for a source language, there is also just one empty item shown in the templates drop down. [TS-6824]
  • All World Server Project Types were retrieved as translation template. This is now corrected to retrieve only the Project Types that are visible for the configured connecting World Server user. [TS-6754]

Installing and Upgrading

  • InstallTool (IT) asks for a username/password combination for every inputparameters.xml file having a <validate> element holding askpasswordtwiceifempty. Out of the box the prompt with confirmation will happen on inputparameters ospassword, issueractorpassword and servicepassword if their <currentvalue> was empty. [TS-6595]
  • Database Upgrade Tool (DBUT) asks for the username/password using masked characters twice to have password confirmation. [TS-6600]
  • After running DatabaseUpgradeTool (DBUT) the InfoShare database jobs were not always picked up the Microsoft SQL Server Agent. The ISH_DisableStandardInfoShareJob.sql and ISH_EnableStandardInfoShareJob.sql files were adapted to communicate with the SQL Server Agent. [TS-6857|TS-7505]
  • Statistics added in the DBUpgradeTool.log file to allow tracking of card id numbers versus sequence numbers. On high-load database systems too much fragmentation could cause future issues. The statistics will look something like:
    InfoShare object information...
    Cards: 330012 [deleted:21 max:387837 sequence:387837]
    LogicalObjects: 76773 [deleted:7]
    VersionObjects: 79971 [deleted:7]
    ContentObjects: 83372 [deleted:7]
    Publications: 16 [deleted:0]
    PublicationsVersions: 53 [deleted:0]
    PublicationOutputs: 84 [deleted:0]
    Baselines: 53 [deleted:0]
    Users: 144 [deleted:0]
    Blobs: 89114 [max:126243 sequence:126245]
  • Normalize the Issuer's certificate thumbprint. When generating the thumbprint of a certificate often the user copies the start of the value from the user interface. This results in a string in notepad that has an invalid character that notepad isn't displaying (Notepad2 and Notepad++ show this as ?) which results in an invalid certificate. [TS-5696]

Database

  • Verified the Standard Edition of Microsoft SQLServer 2012 AlwaysOn Failover Cluster. [TS-8185]
  • Database locking strategy for card locking functions like DocumentObj25 revisited. The earlier 4 attempts with a random wait time between 1 and 10 seconds is now fine tuned to 10 attempts with a random wait time between 500ms (00:00:00.500) and 2000ms (00:00:02.000). The three values can be further tuned by Trisoft_InfoShare_ObjectLockRetries, Trisoft_InfoShare_ObjectLockWaitTimeMinMilliseconds and Trisoft_InfoShare_ObjectLockWaitTimeMaxMilliseconds expressed in TimeSpan syntax. [TS-7561]
  • Oracle job to cleanup unused blobs/pages in ALL_PAGES table only removes 1 blob per execution. To correct the legacy, Database Upgrade Tool (DBUT) will run a script named DBUT_CleanupUnreferencedBlobs.[TS-7969]
  • The job for Oracle logs in the system_message table. The sp_jobClean was tuned to remove message older than 1 year to keep this table under control. [TS-7890]
  • The job for SQL Server databases will run faster. An update statistics of a table containing a varbinary column (e.g. all_pages) takes quite some time. The FullAnalyze.sql has been tweaked to only do full analyzes on tables that don't hold BLOBs. [TS-8009|TS-7959]
  • The Microsoft SQLServer database job triggered by SQL Server Agent is called InfoShareJob. It used to consist out of several steps, and editing the steps required system administrator rights. The job will now contain only one step pointing to stored procedure ISHCore_TrisoftInfoShareJob with a set of optional parameters. The optional parameters allow us to extend the default behavior without requiring a manual system administrator intervention. Out of the box, our job will only do minimal maintenance like rebuilding indexes with more than 10% fragmentation and update statistics without full scan. The name of the job has been matched up in Oracle RDBMs as well for consistency reasons. [TS-6866]
  • Enhancement to Oracle 11g: Increased cache size to improve performance (db_16k_cache_size, db_block_size) [TS-5462]
  • A new language-card-level field FISHLASTMODIFIEDON was added which automatically stores the date of the last modification to a document. The field is initialized with the MODIFIED-ON date when the DatabaseUpgradeTool is run. All client tools are configured to show the FISHLASTMODIFIEDON field under the existing Modified On label through metadataconfig.xml configuration. This resolves the issue [TS-4803] whereby when a user did an autocomplete of the baseline with latest released versions, both the selected version and the previously released version got the same timestamp (when the autocomplete was done).

Change Tracker - Compare

  • Publish with compare fails in ChangeTracker with End tag 'p' does not match the start tag 'fn' (0xC00CE56D/-1072896659) [TS-8425|TS-7237|TS-8587|SRQ-1825]
  • ChangeTracking of a map results in 'There was an error formatting the content. Root element is missing.' while the web client compare works. The <navtitle> tag was not well-formed after comparison. [TS-7627|TS-7638]
  • Publishing with ChangeTracker Compare enabled fails due to a broken merged xml giving error: 'Root Element missing' [TS-6651]
  • ChangeTracker - After text compare, the correction to well-formed xml fails resulting in an error: System.Xml.XmlException: The 'table' start tag on line 3 does not match the end tag of 'tbody'. This may be seen in a table when a row in one column spans multiple rows in another column (for example, column 1 contains 6 rows of information while column 2 is 1 row and spans the 6 rows in column 1). [TS-6264]

Secure Token Service - Authentication and Authorization

  • Certificate update assistance script added to allow updating of relying parties on AD FS v2 and InfoShareSTS like \Setup\STS\InfoShareSTS\Scripts\SDL.LiveContent.Architect-InfoShareSTS-RP-UpdateCertificate.ps1 and \Setup\STS\ADFS\Scripts\SDL.LiveContent.Architect-ADFSv2.0-RP-UpdateCertificate.ps1. [TS-7991|TS-7992]
  • Log in attempts on InfoShareSTS where the username or password contains special characters like &é"'(§è!çà)- failed. [TS-7311]
  • InfoShareSTS has been enriched with basic transformation types based on the relying party prefix. This way the FISHUSERDISPLAYNAME field is added to the claims set for SDL LiveContent Reach and SDL Enrich; and all applications can show the same display name when using InfoShareSTS. [TS-7555|TS-7554]
  • Version numbered the InfoShareSTS internal database files .sdf with a compatibility number. This way when upgrading the just-in-time initialization will work and the typical upgrade scenario will work out of the box. Any extra configuration like relying parties need to be redone anyway for incompatible database files. [TS-7634|TS-7625]
  • The InfoShareSTS is now ready to generate tokens for the Suite by having the claims transformations in place based on the relying party name which is either prefixed with Reach or Enrich. Also the InfoShareSTS scripts for adding a relying party (SDL.LiveContent.Architect-InfoShareSTS-RP-AddRelyingParty.ps1) or a node from a batch server (SDL.LiveContent.Architect-InfoShareSTS-RP-AddNode.ps1) have been adapted and are available in \App\Setup\STS\InfoShareSTS\Scripts\. [TS-6863]
  • The updated InfoShareSTS has received documentation to attempt the following cloud setup: a customer with an on-premises ADFSv2 instance wants to use a LiveContent installation deployed on cloud-based servers outside of their network. All web clients and client tools running from the customers network have access to both their on-premises ADFSv2 and the hosted LiveContent services. The LiveContent services can trust the ADFSv2 as an issuer through certificate thumbprint configuration. The back-end server-to-server communication among LiveContent services can however not contact the on-premises ADFSv2 setup, so they have to be routed differently. The end result is user authentication in the web client and client tools based on your Windows account. [TS-6893]

SDL LiveContent Reach

  • Publishing to SDL LiveContent Reach with a publication title holding characters \ / : * ? ' " < > | & + # . % failed. These characters will now be removed before upload. [TS-8231|SRQ-1587|SRQ-1540]
  • SDL LiveContent Reach introduced an extra call pub.xql?action=prepare_allowed that allows to verify if a scheduled backup is about to start. If so the client code can decide to postpone any write operations. Client code that has been adapted is in publishing (PowerShell scripts), background task synchronizing (PowerShell scripts) and even the browser sync (javascript) when doing a Finish Editing from SDL LiveContent Create. [TS-8460]
  • Publish to SDL LiveContent Reach through Powershell hangs when multiple publishes to SDL LiveContent Reach are running simultaneously. The three-part fix consists out of splitting the logging into a log file per Powershell process based on the process id (e.g. Trisoft.FeedSDLLiveContent.pid1628.log). Making sure that all log writers were closed. And avoiding a Microsoft problem by calling Powershell using extra parameters -inputformat none -outputformat none -file. [TS-7391|TS-7441]
  • Synchronize to SDL LiveContent Reach fails for images having multiple languages (also known as LanguageApplicability) assigned to them. The algorithm would check for language en, de which results in an error like Invalid Language Code. You must use a valid language code such as "en" or "en-GB". [TS-7670|TS-7671]
  • Publishing a publication to LiveContent Reach with Unicode characters in the description field results in errors like Exception while transforming node: Illegal HTML character: decimal 128. The complete support of Unicode also requires a Reach fix identified as LCD-3623. [TS-6952|LCD-3623]
  • SynchronizeToLiveContent events stay at busy state if the topic does not exist in SDL LiveContent Reach. By default the system does multiple synchronize attempts when the configured Reach system is 'temporarily' unavailable; if you don't want synchronize attempts, you should clean the LiveContent url in the Settings area and disable the OnDocStore plugin. [TS-5604]

Other

  • Component Trisoft.Utilities.ServiceReferences.dll has option to override certificate validation through certificateValidationOverride which will now allow logging mismatches. [TS-8307|TS-8279]
  • Version algorithm, the build number system changed. We will now continuously increase the build number. For example version 10.0.312.5 has build number 312 and can be read as March 12 from the year 2013. Version 10.0.1629.5 has build number 1629 and can be read as where every multiple of 1200 indicates a year, so this build was created on 29 April 2014. [TS-6951]
  • JustSystems XMetaL 8 (8.0-CRCL) and JustSystems XMetaL 8 J (8.0-J-CRCL) are added to the server to support single-sourced configuration of the matching AuthoringBridge through synchronization. The matching ClientConfig.xml was extended and also received a new condition name ClientEdition to distinct the regular with the Japanese xml editor as their file name nor version differs.
  • Japanese Search has been enhanced regarding word separation and whitespace usage. Overall Japanese doesn't rely on whitespace for word breaking. In several scenarios Japanese search terms were passed as quoted terms to the search engine resulting in no tokenization and hence less search results. The quoting rules were reviewed and we also added support for the double byte characters for our search operators (e.g. AND, OR,...) and their surrounding whitespace. [TS-6439|TS-6240]