Presently mainly a semantic change, as the dispatch queues are mostly empty, but as we adopt GCD for more things, it’ll be more important to put background tasks in a queue with lower priority than the one handling user requested tasks.
There are crash reports showing the gutter view accessing its datasource in OakDocumentView’s dealloc (based on a “bounds did change” notification fired when releasing a view).
A “Downgrade” button now appear if the version on the server is lower than the current one. This would e.g. happen if the user has upgraded to a nightly build and does a manual version check (for non-nightly builds).
If the version on the server is the same version as the one running, the user can make a “Redownload” button appear by holding down shift (⇧) when running the update check.
We now explicitly disable it for targets that hasn’t yet been upgraded to ARC. This way, it’s easier to get an overview of which targets hasn’t yet been upgraded and ensures new targets has ARC enabled.
This is only motivated by easier transition to ARC (one framework at a time) as the ns/attr_string.h header contains manual memory management, which thus cannot be included by a framework using ARC, yet the code needs to be in the header since the code is template-based.
The help indexer requires the pages to be valid XML and (silently) fails to generate the proper index (with anchors) when they are not.
This fixes issue #548.
Zoom is an OS X accessibility feature that lets sighted visually
impaired users magnify screen content. This commit notifies Zoom of
changes of TextMate’s caret position on screen so that Zoom can
automatically follow the caret when the user moves it.
This commit adds support for most common accessibility
methods/attributes to the OakTextView component. In user's language,
VoiceOver users (i.e. blind and visually impaired users) can now use
TextMate to read and write text.
A few less used accessibility attributes remain to be supported -
especially AttributedStringForRange and StyleRangeForIndex which should
allow blind users to e.g. see and seek for spelling errors in text.
This avoids depending on the user having installed the JSON gem (and fixes problems where user has it installed, but for a different ruby than the one used to run this script).
This is when generating HTML for the release notes, manual, and, where we actually need this, list of contributions.
Long-term we should probably switch to template tags as we are effectively adding code to the (otherwise declarative) build graph, which means we don’t have any way to tell if the generated HTML is up-to-date or not (as that would require analyzing the embedded ruby code).
