| Age | Commit message (Collapse) | Author |
|---|
|
Also fixing glitch with search input field where pressing Return would insert newline.
|
|
Memory used for RAM storage of media along with navigation history so it can be restored instantly.
|
|
Opening an URL or making a search query from the command line.
IssueID #232
|
|
Several regressions occurred when the split view mode was implemented.
|
|
|
|
The primary purpose is to aid navigation in split view, so one can see exactly which links have been opened.
|
|
The serialized state version was bumped, so old states are incompatible and are ignored.
There were redundant events posted during launch.
|
|
Root focus switching and opening links in the other root.
|
|
Various refactorings and fixes to handle root-global and window-global state, root-specific palettes, and proper coordinate system changes (e.g., when opening menus).
UI events are posted and handled in the context of a specific root.
|
|
Most actions should occur in the context of the current UI root.
|
|
|
|
Notify when hover state should be updated again after scrolling.
|
|
Added a new mechanism to issue periodic but not per-frame commands. This is used for main-thread operations like checking if it's time to fade away the scrollbars.
Scrollbars are faded away completely on Apple platforms. Adjusted list right margins accordingly.
|
|
Previously glyph caching was done during text rendering, but that would mean lots of swapping between rendering stuff to the cache and then immediately afterward rendering to the display. There would be a swap per each new glyph.
Now the entire document's glyphs are precached at once when a request is finished. Glyphs are also cached in larger batches when new text needs to be drawn.
|
|
IssueID #178
|
|
|
|
Phone mode uses a modified user interface. Work in progress...
|
|
Any text that doesn't look like a URL is passed onto the configured search URL as a query string.
IssueID #157
|
|
Rather than simply limiting each tab's cache to 50 most recent URLs, there is now a user-configurable maximum size. If more content is cached, the oldest/largest responses will be removed from memory.
The default maximum cache size is 10 MB.
IssueID #109
|
|
|
|
It appears at least on macOS, SDL is doing a while lot of stuff while waiting for new events. Perhaps because it has some sort of high-frequency input/sensor processing? Not sure.
Now Lagrange will idle by polling events every 15 ms and sleeping in between. This reduces CPU time by an order of magnitude on macOS. Need to still test on other platforms.
|
|
This is a very powerful mechanism: translate the contents of any request to something else when the original MIME type matches a configured regexp.
The external hook command may still elect not to process the request.
|
|
|
|
IssueID #25
|
|
The DocumentWidget was not actually deleted when a tab was closed, only hidden.
|
|
|
|
|
|
|
|
|
|
Improved drop and drop event handling: multiple dropped files/URLs open in new tabs. The application registers gemini: as a handled URL scheme.
|
|
|
|
|
|
Even preformatted lines may need to be wrapped so the content remains visible, since there is no horizontal scrolling. However, this is off by default so ASCII art isn't broken in narrow windows.
|
|
Make requests via a proxy.
TODO: What about the server domain verification? Check against the proxy
hostname?
|
|
Finder is accessed via AppleScript.
|
|
|
|
|
|
|
|
|
|
Fonts are shared by all documents, so zoom likewise needs to be
app-wide.
|
|
|
|
|
|
GmRequest is given a pointer to the GmCerts to use. Also, timeouts are handled via a posted command so they get processed in the main thread.
|
|
|
|
App maintains the visited URLs database, but each DocumentWidget has its own stack of recent URLs for timeline navigation.
|
|
On macOS and Linux.
|
|
|
|
|
|
|
|
|
