“Lightweight Software #Transactions for Games”, a #paper from 02009 about SpaceWars3D, where they tried to “realize the #performance potential of multiple cores” with #transactional-memory, only to be terribly disappointed. “[W]e depart from classic STM designs and propose a programming model that uses long-running, abort-free transactions that rely on user specifications to avoid or resolve conflicts.” #videogames
on 02026-09-1810 years ago Mozilla changed #IndexedDB to have non-durable #transactions by default, adding a “readwriteflush” mode for durable transactions. #HTML5
on 02025-12-27Disqus hasn’t yet lost Jonas Sicking’s comment about the synchronous API to #HTML5 #IndexedDB: “The reason we’re not permitting access to the sync version in the main thread, i.e. in Window environments, is that synchronous IO is terrible for performance. This would remain the same no matter what syntax is used, so would not be affected by a switch to a module based syntax.
However any environment that isn’t the main thread should be able to use the sync version. That isn’t affected by weather we use module syntax or not though. We could allow access to the synchronous API to any non-main thread environment. Currently the only such environment is web workers. If more are added in the future we might be able to add them there too, independent of if we use modules or not.” He also comments on how their #transactions work.
on 02025-12-27Jonas Sicking (an #HTML5 #IndexedDB spec author) explains how its #transactions work; normally they’re pretty transparent, and you can, like, read some data and then write some data in the callback when the reads succeed.
on 02025-12-27#HTML5 #IndexedDB #transactions commit implicitly when you return to the event loop without any pending requests, which seems error-prone
on 02025-12-27suggestions for running #CICS for #transactions in #emulation. “it depends on the installed operating system, if you manage to get hold of a copy of z/OS, then it's accurate. most people run the old MVS3.8 system that was the last release that was out there in the open without restriction.”
on 02025-11-11410-page #PDF #documentation for #CICS explaining how its #transactions work
on 02025-11-11I think "Kicks" is a #CICS clone or fork? For playing with #transactions on your IBM 360 in #emulation I guess. “KICKS is an enhancement for CMS & TSO on IBM mainframes (or emulators) that lets you run your CICS applications directly instead of having to ‘install’ those apps in CICS. You don’t even need CICS itself installed on your mainframe. KICKS provides a high level of source code compatibility with CICS. You can migrate applications either way between CICS and KICKS simply by recompiling.”
on 02025-11-11#PDF #paper “Virtualizing Transactional Memory” by Herlihy #transactional-memory #transactions #toread
on 02025-11-11“Does Transactional Memory Keep Its Promises? Results from an Empirical Study” #PDF #paper #transactional-memory #transactions #toread
on 02025-11-11#PDF #paper “Software Transactional Memory: Worst Case Execution Time Analysis” #real-time #transactional-memory #paper #toread on #transactions
on 02025-11-11“Lazy Irrevocability for Best-Effort Transactional Memory Systems” #PDF #paper #toread #transactional-memory #transactions
on 02025-11-11#PDF “Making the Fast Case Common and the Uncommon Case Simple in Unbounded Transactional Memory” #paper #toread #transactions #transactional-memory
on 02025-11-11