Activity for Mark R. Rubin
-
Mark R. Rubin committed [178242] on Rosegarden - Code (git)
2023-04-02: MIDI file import fix
-
2023-03-30: Minuscule update -- missing version bump, README.md fixes
-
2023-03-29 ChordNameRuler/AnalysisTypes/ConflictingKeyChanges refactoring
-
Retrograde and Retrograde Invert commands broken
-
Manually restore master to being exact copy of upstream/master
-
2022-12-31: Pseudo-merge with rosegarden-official master [fcb8f6]
-
Merge branch 'master' of https://git.code.sf.net/p/rosegarden/git
-
I agree that the solution lies in fixing RG's logging. The problems with that subsystem go all the way down to the fact that RG_WARNING (which should be, as per its name, for warnings) (regardless of how they might be output or not) is used throughout the code for debugging instead of RG_DEBUG or some similar, correctly-named macro/class/method/whatever. I disagree that (true) warnings shouldn't be accessible to users. Remember, I proposed that the info window would only be displayed on user command...
-
2022-12-29: Matrix editor pitch note color
-
Info window for warning messages, etc.
-
2022-12-21: Pseudo-merge rosegarden-official master [51d363]
-
Merge branch 'master' [51d363] of https://git.code.sf.net/p/rosegarden/git
-
2022-12-14: Matrix Editor Panner fit notes
-
Branch v0.5_merge_request_15
-
git-restore to current v0.7 master
-
Revert master to latest upstream v0.7 5530160c6
-
Merge branch 'master' of https://git.code.sf.net/p/stm32flash/code
-
Merge branch 'master' of https://git.code.sf.net/p/stm32flash/code
-
Mandatory clean local repo prior to `git revert upstream/master`
-
Revert to last official, pre-thanks4opensrc, 2020 0.5 commit
-
Thanks, @tormod. I'm not set up to test this right now, having been stuck working on a non-embedded open source software project for most of this year -- https://github.com/thanks4opensource/rosegarden-fork, from https://sourceforge.net/p/rosegarden/git/ci/master/tree/ -- but I hope to get back to using STM32 MCUs again soon and will try v0.7 then. I did review the changed code and it looks good to me. Thanks for incorporating my specific needs in the mainstream release, particularly with this improved,...
-
2022-12-01: Chord/Key Ruler enhancements, and extant key display
-
2022-11-20: Main Window Chord Name Ruler bug fix
-
thanks4opensource / thanks4opensrc fork
-
thanks4opensource / thanks4opensrc fork
-
Chord Name Ruler rewrite and Matrix Editor note label enhancements
-
An improvement, but still doesn't have ... Option to stop playback in middle of loop and restart without jumping back to start of loop. Ability to play loop exactly once per command, in addition to endless repeats. Ability to adjust loop start/end points without redrawing entire loop from scratch (much better control of long loops with spans greater than visible panned/zoomed range). Visual feedback on where loop is in composition if not currently pan/zoom visible. All looping capabilities applicable...
-
Chord name ruler
-
Chord Name Ruler rewrite and Matrix Editor note label enhancements
-
Chord Name Ruler rewrite and Matrix Editor note label enhancements
-
Merge branch 'master' of https://git.code.sf.net/p/rosegarden/git
-
Chord Name Ruler rewrite and Matrix Editor note label enhancements
-
Updating this post. My original claim was correct: [cd3d4b] is the correct commit (it is master [eda670] merged into my [942d1e]). It was pushed here successfully and shows as "Tree [cd3d4b] thanks4opensrc_devel /" at https://sourceforge.net/u/thanks4opensrc/rosegarden/ci/thanks4opensrc_devel/tree/ But this merge request, made after the push and with thanks4opensrc_devel as the source, only has [942d1e]. Clicking "Refresh Commits", above, updates this page but leaves [942d1e] unchanged. This is now...
-
I will be beyond amazed if, 15.8 years later, the original poster or anyone else is still subscribed to and/or interested in this ticket, but on the extremely unlikely chance that is so, please see Discussion at Feature Request #504 and/or Merge Request #78. Comments, feedback, and particularly bug reports welcome.
-
I have submitted Merge Request #78 for commit 942d1e. It represents a major enhancement/redesign/rewrite of the Matrix Editor tools, including a rethinking of the original proposal in this feature request, along with the incorporation of ideas from Feature Request #295. Almost all of the current tools and behaviors are retained despite now being largely duplicates of new features that supersede them. From the merge request and commit message ... The intent of this large merge is to improve user workflow...
-
Apologies, my lack of git-knowledge. [cd3d4b] is hash in my local repo, [942d1e] as generated by the merge request seems to be correct (but still is not showing up at https://sourceforge.net/u/thanks4opensrc/rosegarden/commit_browser).
-
Feature Requests #504 and #295: Matrix Editor tools rewrite
-
Merge branch 'master' of https://git.code.sf.net/p/rosegarden/git
-
Feature Requests #504 and #295: Matrix Editor tools rewrite
-
Feature Requests #504 and #295: Matrix Editor tools rewrite
-
Merge branch thanks4opensrc_fr508_marker_ruler[f36420] into thanks4opensrc_devel[fc714d]
-
Looks good to me. Using the updated procedure (same as original but make sure mouse is not in NotationView area), get repeatable crash with master/[1979b2], none with bug1644/[e67b14]. Not that it matters, but I'm speculating that this wasn't reproducible for Philip because he's on a Qt6 (??) vs my (and Ted's?) Qt5. And that the bug was real but an edge case, and 6 is either insensitive to it or specifically checks for and avoids it.
-
Yes, you're correct. My simplistic check of "news" vs "deletes" wasn't taking into account the delete before makeNote(). As I said, it doesn't matter anyway due to the fact that Qt cleans up orphaned QGraphicsItems. And also yes, doing a delete of a null pointer has been safe in C++, either since the beginning or at least for a very long time.
-
I don't see much difference between being in "Draw Notes and Rests(F3)" vs "Select and Edit(F2)" mode when the undo of the segments (Ctrl+Z) is done. I do see a difference -- I don't think it ever crashes, although I'm not 100% sure -- if the mouse is not in the staff view area during undo. I added more debug and don't see any double frees of NotationStaff::m_previewItem, although they're definitely being leaked: NotationStaff::showPreviewNote() created 0x3bee4b0 NotationStaff::showPreviewNote()...
-
Yes, this is one is very hard to track down. After looking at the code and comments in [cf0a28] it seemed possible that a double free was happening, so I added debug output (in my current working tree). While the crash is 100% reproducible on my system (in commits ranging across my own feature branch, master 22.06, and others), with debug sometimes it crashes, although not with any double free that I can find, and sometimes not. I haven't looked into the subject for many years, but always assumed...
-
Crash in notation editor when undoing create segment, incl. 22.06
-
Crash when recording
-
(I posted a response here that was intended for a bug report. Apologies for the mistake and the spam.)
-
Philip Leishman wrote: Maybe we are just going to have to agree to disagree !! Yes, that's probably the bottom line. The following are merely details ... ell1e wrote: there should be no unnecessary jumping In principle I agree, the difference here being the "unnecessary" part. Beating the dead horse of what I've already said, if I've defined and activated a loop, I want to hear that loop. And in my version it's very obvious when a loop is defined/active, even if that loop's time range is scrolled...
-
Philip Leishman wrote: Maybe we are just going to have to agree to disagree !! Yes, that's probably the bottom line. The following are merely details ... ell1e wrote: there should be no unnecessary jumping In principle I agree, the difference here being the "unnecessary" part. Beating the dead horse of what I've already said, if I've defined and activated a loop, I want to hear that loop. And in my version it's very obvious when a loop is defined/active, even if that loop's time range is scrolled...
-
Typo in merge request above, should be "as per Merge Request #74". Link was correct, text had wrong request number.
-
Feature Reqest #508: Marker ruler enhancements and fixes
-
Feature Request #508: Marker Ruler fixes and improvements
-
Merge branch 'thanks4opensrc_bug1605_playback_loop' into marker_ruler
-
Marker ruler enhancements and fixes
-
MIDI file import/open crashes
-
Thanks for the quick feedback, Philip. I was going to post something about my latest commit but you beat me to it. Responses to your comments below, but first a recap on what I've submitted: The looping interface is split into two separate parts: Whether a loop range has been defined and is currently active (displayed in the loop ruler), and whether looping/repeating is on (the loop button(s) are clicked into the "on" state). Playback behavior applies, in the same way, to the loop range if defined...
-
If and when reviewing/merging/cherry-picking/etc please use updated [9ea98d]. For description of changes see [0d71f8] commit message (pre-merge with latest master [63c7e4]).
-
Merge branch 'master' 2af082 into thanks4opensrc_bug1605_playback_loop 0d71f8
-
Merge branch 'master' of https://git.code.sf.net/p/rosegarden/git
-
W.I.P. marker_ruler
-
Bug #1605 -- Looping UI fixes and new features
-
The following is a response to a post on rosegarden-user, here instead of on the mailing list as was requested there. As always, these are my opinions and contributions and may be disputed, rejected, and/or ignored by any and all as they so choose. On 6/20/22 5:44 AM, Ted Felix wrote: On 5/21/22 5:09 AM, Lorenzo Sutton wrote: So as I understand correctly if advanced looping is on in preferences selecting the loop toggle in the transport will loop from start to the end of the last segment unless a...
-
An initial response. More to come when I commit the latest fixes and improvements to my branch. Regarding the test issue: I am a little concerned about your problem with the reference_segment test. I can compile and run it OK. Can you specify the problem ? Yes. The latest test/reference_segment.cpp fails compilation on my Qt 5.9.7 system: .../rosegarden-git/test/reference_segment.cpp:119: undefined reference to `bool QTest::qCompare<long, int>(long const&, int const&, char const*, char const*, char...
-
Thanks for finding these problems, Philip. Please accept my apologies for breaking test_notationview_selection. I hadn't looked into the unittest code and had gotten out of the habit of building and running it because I didn't think it tested GUI interactions like the changes in this merge request. Totally my mistake. The tests are great and many more of them are needed, although I still don't see how non-button actions, like rectangle-select in the editors, could easily be simulated. Of course the...
-
If and when merging please use updated [baf448] which has a bug fix and additional enhancements (see commit message).
-
Bug #1605 -- Looping UI fixes and new features
-
The merge commit should be the [c834b1] HEAD of thanks4opensrc_bug1605_playback_loop, not [48d4e3] which was before the merge with master [f2516f] aka 22.06. I don't know why SourceForge gets the wrong commit, both when the merge request was created and when "Refresh Commits" is clicked now.
-
Looping mode fixes and improvements (Bug #1605)
-
Bug #1605 -- Looping UI fixes and new features
-
Update to 21.12 master [f2516f824]
-
Bug #1605 -- Looping UI fixes and new features
-
Merge branch 'master' of https://git.code.sf.net/p/rosegarden/git
-
Merged master [1c2e80] into thanks4opensrc_devel [63b543]
-
Matrix Editor graphics improvements, BUG#1153, and WIP FR#504
-
Merge branch 'thanks4opensrc_fr504_matrix_editor_change_active_segment' [22ebff] back into thanks4opensrc_devel [94b71a]
-
Crash when resizing segment
-
That's a good example of why the octave number is semi-meaningless. The ones I was thinking about were canonical guitar chords, which often have repeated root notes such as root-fifth-root-third-fifth-root or root-third-fifth-root-third-root ("E" and "G" open chords respectively, or the same voicings transposed to different roots in the "CAGED" system of barre chords). And it's not just guitar -- multiple roots are common in two-handed piano playing and ensemble/orchestra scores. Which one of the...
-
Thanks, Yves. This really helps. At first I was surprised to find you had completely removed the octave number instead of fixing it to be correct. But I thought about it, and can see the logic. A user can always look at the notes (while working in either the Notation or Matrix Editor) to see the octave, so except for a very MIDI-oriented mindset, it's redundant. I still think AnalysisTypes.cpp could use some more work in the future. If nothing else, there's things like: // This is an ugly loop. Is...
-
The new RG_WARNING doesn't automatically insert spaces. Is this really a problem? For me, not particularly. Like I said, I find the qDebug/qWarning family's automatic spacing and newlines both convenient (as per its obvious design intent) and limiting (can't concatenate strings or do multiple outputs to the same line). The true question is what is RG_WARNING for? To write short messages (including QStrings) or to write long ones including the dump of QtObjects? That is the question. I've been using...
-
Yves: Sorry, but I think there's a problem with this. The current RG_WARNING built on top of QDebug() has many of that class's features that I have mixed feelings about, including the automatic addition of spaces between inserted objects and the addition of the newline at the end -- all very convenient, but that prevent concatenating text or appending to previous output when desired. But more importantly, all(??) Qt classes implement some kind of QDebug &operator<<(QDebug &stream, const SomeQtObject...
-
Thanks, Yves. Those are interesting workarounds which I wouldn't have thought of. And in the example I was using (four_part_harmony_stems.png) there aren't any rests so the worst of the problems go away. Editing two coincident segments on the same track and staff works but is a little clumsy. One of the nicest things about working with multiple segments/tracks in Notation is that moving the mouse from one staff to another automatically changes the active segment without having to use the scrollwheel,...
-
I would consider all RG_WARNINGs as essentially output for developers ... Point taken, and if that's the intent and the code follows it, all's good. I know I've seen counterexamples, e.g. MatrixView::slotDiatonicTranspose() if (intervalDialog.getChangeKey()) { RG_WARNING << "Transposing changing keys is not currently supported on selections"; } which returns immediately after that, doing nothing and potentially leaving the user going, "What? Nothing happened. Did I do something wrong?" But yes, most...
-
My newbie's 2 cents: I'm all for anything that makes logging more configurable, at runtime or compile time. I've found little use for RG_DEBUG because turning it on, either via -DCMAKE_BUILD_TYPE=Debug or manually editing Debug.h (I wish there was a -D flag to going to a ?= variable in a Makefile, but I still don't understand cmake) results in such a spew of messages that I can't find my own output among them. So I just use temporary RG_WARNINGs in my code, or even fall back to a raw qDebug() sometimes....
-
Thanks for pointing this out. I've been meaning to figure out why the ruler shows up sometimes and not at others but have been too busy with other self-assigned tasks. The other obvious bug is that the chords always show "0" for the octave. I did take the time to look into that, and the code is both clearly wrong and easily fixable. But again it went onto my personal "to do" list. And if I do get around to it, I'm likely to add a whole bunch of "Feature Request-y" stuff, like more chords (sus2 and...
-
I'm very late to this party, but now learning about the "Advanced Looping (experimental)" preference ("Nobody reads the release notes" -- thanks4opensrc, 2022) it's a huge improvement. I've never been able to imagine anyone having use for Rosegarden's continued playback to "infinity and beyond" after the last segment's end, and it always drove me crazy. One more thing that I've been thinking about Feature Requesting but will piggyback here instead: Why does playback, when a loop range is set, jump...
-
Three half-notes in 4/4 measure
-
Small fixes for Bug #1153
-
There were still some small bugs remaining in EventFilterDialog.cpp, even after the last significant changes made in 2009 and 2013: The "shortest" Duration value was being incorrectly extracted from the ComboBox, effectively being "longest" and thus clearing the selection instead of including the desired pitches/velocities/duration. Default "lowest" and "highest" durations were reversed, again resulting in nothing passing the filter by default. I've patched the problems and made some other small...
-
Small fixes to EventFilterDialog, Bug #1153 2022 update
-
Merge branch 'master' [6a0662] of https://git.code.sf.net/p/rosegarden/git
-
Merge Feature Request #507
-
Feature request #507: "One-click" colors for multiple segments
-
W.I.P. color segments, Edit -> Segment colors
-
Merge branch 'wip_colorize_segments' into thanks4opensrc_devel
-
Fixed missing Matrix Editor "Add Control Ruler" menu items
-
"One-click" colors for multiple segment
-
Thanks, Philip. Tested and didn't get any crashes.
-
Philip Leishman wrote: Very complex code !!! Yes, I've been noticing some of that in RG. :( Thanks for the fix. I've tested and didn't see any problems even when torturing it with clusters of notes on both sides of the staff, sharps/flats, beams, etc. I can't guarantee I tried everything but what I did couldn't break it. :)
-
12 years later and this bug has returned from the dead: Create segment. Notation editor. Add note (required, no crash if not). Add Layer. Undo. (gdb) where 4 #0 0x000000000086eeae in Rosegarden::Composition::getBarNumber(long) const () #1 0x000000000086f8c9 in Rosegarden::Composition::getBarRangeForTime(long) const () #2 0x000000000060c5a7 in Rosegarden::NotationStaff::getEndTime() const () #3 0x000000000064f5b8 in Rosegarden::NotationScene::getCursorCoordinates(long) const () (More stack frames...
1 >