Activity for Michael Martin
-
Michael Martin committed [d6583f] on Git Repository
Fix credit
-
Fix Mac options for linker maps
-
Move the git conversion scripts somewhere more out-of-the-way
-
Minimum Viable Tag Monster is now complete.
-
Convert tags to tags or branches as appropriate
-
Better links from orphan branches to their releases
-
Fix checkout root when converting testprogs/techdocs
-
Correctly handle branches that didn't copy from the right start point
-
Handle the two full branches where branch-copies were missed by git-svn
-
First draft of SVN->Git branch migration.
-
Tag Monster can recreate the release history and tag each release now
-
Incremental updates for the tag_monster script
-
Produce synthetic histories for branches and releases in the tag monster
-
Major refactoring/commenting pass on Tag Monster script
-
First draft of the "Tag Monster" script.
-
Also clone and update git versions of testprogs and techdocs
-
Centralized git migration target locally
-
Incremental Git update
-
Much faster author extraction
-
First draft of an SVN-to-Git converter for VICE.
-
Fix input of trans-ASCII Unicode characters under SDL2
-
This is more general than that; changing the drive type to something image-incompatible (like 1541 to 1581) also doesn't detach any images and commands like LOAD "$",8 will start misbehaving. I think we should detach on any drive type change.
-
Something very like this was actually the first draft. :) Unfortunately,_fullpath does not meet our full requirements for this function. Most of VICE assumes a case-sensitive filesystem, and that means that for case-preserving/case-insensitive filesystems like on Windows and Mac, we need consistent capitalization of the filenames no matter what the user types. _fullpath and GetFullPathName both preserve the casing that was typed in, which doesn't help for command line arguments. In a dramatic irregularity,...
-
The same disk image can be attached to multiple drives
-
This has a best-effort fix in the latest release. If you do some really aggressive shenanigans with hardlinks and such you can probably still defeat it, but the current level should be good enough to prevent simple mistakes,
-
Handle CSCP filesystems on windows. realpath(3) covers this on Mac.
-
Canonicalize path before testing. Still needs some work on Windows.
-
Add a new archdep function to wrap realpath(3)
-
Pop error messages if a mount/autostart failed
-
ui_error() does not have a window caption argument
-
Fix #1468: Do not permit the same image to be mounted simultaneously across
-
VICE: Loading snapshots result in screen getting cut off
-
I've done some experimentation on historical versions and I believe that dqh fixed this in r40623.
-
The same disk image can be attached to multiple drives
-
Fix typo in installsome()
-
Updated fix for Bug #934.
-
Upgrade internal libmikmod to 3.3.11.1.
-
Mark "master" branch as deprecated
-
Update libmikmod documentation
-
Remove dll export/import logic from internal mikmod (Fix Windows Build)
-
Upgrade internal libmikmod to 3.3.11.1, for Aarch64 support
-
Update Win32 installer generation instructions
-
Correctly handle Unicode Replacement Characters
-
Put the Text Input code in the correct SDL version branch
-
Don't leave behind the updated Remix Pack 4 on uninstall
-
Fix misplaced FlashRect when renaming Captain/Ship
-
Restrict SDL_Start/StopTextInput calls to the UI thread
-
Remove beta marker from version strings
-
Remove outdated specific trademark acknowledgement
-
Don't try to relink frameworks through symlinks
-
SDL1 Hardware Scaling Goes Nuts
-
After some more experimentation, I've concluded that this is an upstream bug: the nonsense I receive -- or the points where the viewport just stops resizing -- are data pulled directly out of the SDL resize event itself (or the resize event simply stops firing altogether), so VICE never actually gets a request to rework the display.
-
Update Win32 installer metadata for updated Remix Pack 4
-
Fix two bugs regarding stale SIS messages, from Kruzen
-
Fix the ZFP dialog desync bug introduced post 0.7.
-
Update keyjam.exe to use SDL2
-
Fix a render crash on Intel GPUs in 32-bit DX9
-
Rework SDL2 render driver selection.
-
Create a draft 'exec forwarder' to let us transparently hide x64sc.exe et al under bin
-
Updated Win32 and packaging documentation
-
Deeper automation of the Win32 installer
-
Update the build instructions for all platforms.
-
Fix strncpy warning with correct bounding value
-
Fix oscilloscope display under SDL2
-
Remove potential format-string attack on log_displayBox on macOS
-
Fix an alarming bug in AlarmTime_compare
-
dirname can modify its argument; don't pass in a const char *
-
Fix strncpy warning by making it more obviously dangerous
-
Defensive casting to silence clang warnings
-
Fix SaveSummary return type
-
Fix icon on Windows with MinGW/MSYS2 builds, from Ala-lala
-
Update Changelog, WhatsNew, and Authors to match progress
-
Indentation and clause order refinements on NSIS installer
-
Update in-game version to beta status
-
Rewrite the NSIS installer logic to use LogicLib
-
Update installer version to 0.8.0b
-
Remove obsolete NSISdl autodownload code from Win32 installer
-
Automatically generate the DLL dependencies of the MSYS2 UQM build for installation and uninstallation
-
Switch Win32 Installer download plugin to InetC, which supports redirection and HTTPS
-
Rework the packages.nsh generation for v0.8.0 and MSYS2
-
Improve procpkgs.sh to remove bashisms and shellcheck lints
-
Merge shared code in VIC-IIsc's set_sync_factor function
-
crash on "restore settings" when NTSC was active
-
Fixed in r37776, including x64-linking fixes.
-
Fix #1220. Ensure that changing MachineVideoStandard doesn't
-
This is my first attempt at a patch for this. I'm trying to keep configuration from leaking between systems and chips, so I've expanded the VICIIsc resources and then had the C64 resources rely on those. That might break x64, but (a) we don't build that and (b) x128 and x64dtv seem to work fine.
-
I've found the cause of the problem, but don't have a patch yet. The issue here is that setting the MachineVideoStandard resource assumes that the VICIIModel resource has already been set to a reasonable value. When setting the model directly, or during machine initialization, this is guaranteed, but the resources_set_defaults function just blindly iterates through all resources in registration order and does them backwards. This results in a call to raster_draw_buffer_alloc being skipped when it...
-
Implement archdep_file_exists for Windows
-
Remove unnecessary Unix check on c1541
-
Update and reorganize credits for 0.8.0.
-
Add E1 to pt13.fon so credits correctly render András Barják's name
-
Remove redundant statefile saver from new format
-
Merge commit '7326ff096d0fc60086f2' into v0.8.0
-
Spelling and punctuation pass (bug #1200)
-
New, fuller manpage (bug #1204)
-
Update version number to 0.8.0
-
Do not use the D3D9 SDL2 driver if any alternatives are to hand.
-
Fix Linux build; regcomp plays poorly with symbol detection for some reason
-
Fix MinGW build; check for whether we must explicitly link -lregex
-
Update version to 0.7.2
1 >