BUG You should be able to click on a note to select it, then start dragging it around. Likewise clefs, and probably myriad other things. This is all broken too. [This looks like it's probably related to a couple fixes ago, my hack that said PLEASE REVIEW! on it. It's more bogus w and h variables, seems like to me. I haven't given up any hope of sorting this one, but it's not the best use of my time if there's any remote possibility Chris might steal enough time to have a look himself.]
Observation Double clicking on the document title and so on in multi-page view should open the document properties editor. (Changed from bug to observation because this really isn't like any of the other problems I'm chasing. Like time signatures, these things just aren't under the same umbrella, and this inconsistency is going to have to stay for the foreseeable future.)
Observation Sweep selecting in a multi-staff environment is a bit delicate, and the track header indicator of which staff is active is very touchy. When selecting around the divide between two staffs, I frequently have to break away and come at the selection from a different angle to get the right notes, and the active staff indicator usually trips and indicates a different staff from the one where the selection is, so it doesn't seem to be a very accurate indicator.
BUG I'm in select mode (arrow tool) and the duration monobar is in rest mode. I want to insert a note. I click the note/rest switcher button to switch the palette, expecting it to switch the palette to notes and also switch the tool to the pencil, but this does not happen. I've been trying to ignore and get used to this fact, but it keeps irritating me, and if I keep being irritated by what I expect to happen not happening, I think it means the interface is wrong, and it should be made to behave as I keep expecting intuitively.
BUG The status bar is still completely screwy. The most important thing missing for me is the little thing that shows what pitch you're hovering over. I miss that. I miss that too. -Jani
BUG Add a key change, try to select the key signature, and it disappears! WTF? It seems to continue to exist and to have an effect, but it's quite invisible. Strangely enough, in a basic one bar test segment, right clicking the whole rest and “make invisible” makes the key signature reappear briefly. Also, when triple-clicking to select the whole staff, the key signature goes blue briefly before disappearing, so it's not something completely stupid like trying to paint it in NPF with the selected color goes wrong and draws it transparent. Although that may yet turn out to be the key. (Transparency brought on by NPF, I mean.)
BUG Shift+click+drag to micro-position does not work
Options dialog looks OK from here, but I didn't actually try this on this run. It has worked previously, and no surprises are expected here.
Likewise not tested this run, but it has worked in the past.
Option dialog looks OK. Did not actually try different options to see if there are any problems here. (Likely if there are, they're old, previously-documented legacy bugs anyway.)
Works (to my surprise, actually).
Works, with caveats. (Ignore the caveats for Thorn.)
Event filter layout looks good, and I tested a handful of different “pick this but not that” scenarios with consistent results. (It reminds me how this thing has never been as useful as it ought to have been, and needs a rewrite one day.)
Works, but does not open to Notation preferences.
Observation Having this open to the notation page is involved. There's no existing hook in the IconStackedWidget for setting it to something from the outside, and it's not clear at a glance what to do to make it so. It was only a cursory glance, but it looked complicated enough I dumped the entire patch and decided to quit before it became a black hole for time over something trivial.
Some of the font mappings have problems here. Particularly Inkpen. I don't think I care much, and now that we don't do printing in house, none of this matters all that much anyway. (If LilyPond can use different fonts, that's something we should support. Especially if it can use that stupid jazz fakebook font some people are so fond of, which is the main complaint we've gotten over the years.)
In my multi-staff test view, none of the radio buttons are checked initially, and the size is set to 6. Changing the size via the menu or the more accessible toolbar control gets the menu updated. Too minor to even flag as a bug, since the menu version of the control will almost never be used anyway. The basic functionality works fine.
There are some spacing issues with big runs of 64th notes, and changing the spacing with this control does not resolve the issues as I would expect, but it seems to be basically functional, and all of this is probably down to “layout in Thorn sucks compared to Classic” generally.
Lyric editor layout looks OK. Did not test. I've always hated the lyric editor, and can barely manage to use it on a good day, so how would I ever spot subtle problems?
I'm not going to go through all of this crap point for point. I think there may be problems with initial settings not taking effect and/or settings amnesia, but it's all so minor I just don't care at this point. Fry big fish first.
Tempo dialog looks good generally. Tap to set tempo works.
Inserting a tempo from the dialog doesn't seem to behave predictably, and neither do the tempo ruler functions. This needs re-examination in a cleaner test environment than I have now.
Too much space between numerator and denominator. Took a whack at that one, and any improvement is so subtle I'd have to compare before and after screenshots to see if I shaved any pixels. Stubbornly resistant to change. Something in numBox and denomBox has a large sizeHint() apparently. Even changing to QSizePolicy::Maximum had no effect on the vertical spacing, although it did wreck the horizontal spacing nicely.
Clef dialog has too much space around the BigArrowButtons. Maybe that's the culprit in the time sig dialog too. Maybe the whole thing needs to be redone (and the time sig too) on a grid layout, instead of the VBox full of HBox chunks I wrote for both of these to sort out the previous layout woes. I think we can live with it like this and move on though.
Observation: This whole ”<inexact>” for duration is dumber than a box of rocks. Unless you want the segment to be the exact duration of some note, which is incredibly rare, that whole control is pointless and stupid looking.
Did not test well. I don't understand interval theory that well, etc. and there are a lot of choices to test out. I think it's probably OK though. GUI blacked out for a long time after executing a random transposition with random options, but no crash. (Still, after two days of continuous running on this review, creating a god-awful mess out of the poor composition along the way too.)
Did not test, but I've actually used this function for real recently, and it worked then. This function ROCKS!
Observation: The dialog layout is a bit too wide. It could be narrower without negative consequences, and it would have better visual balance, but I really don't care enough to even think about it right now.
Did not test, but I have tested extensively in the past, and do not expect surprises. I added some new marks, and the remove all marks icon, and tested in the course of that. (Of course you never expect surprises, but dammit man that's a lot of thing to double check.)
Layout looks OK. Did not test today, but I tested this some when fooling with the new tr indication, and I think it's generally working. (Would not worry much about a release over this one way or the other, as it is a cool feature but little used, and it might easily go two releases fully broken before anybody noticed.)
Works well enough for me. Tested an arbitrary fingering too. Everything seems as good as it ever was here, and that's good enough for now.
Surprisingly enough, a simple test worked. Set one note each to Cross, Mensural, and Triangle. No crash, and it works as well as it ever did, which is to say not very well with respect to head placement relative to stems, and beams. But it didn't crash. That's surprising. It crashed last time I fooled with it, and I'm not going to try too hard while I'm on a roll here.
Tried several respell options. Tried cautionary and cancel cautionary. Not tested exhaustively but it seems solid.
Selecting a group of notes and un-beaming then auto-beaming them leaves them un-beamed. Not sure if this is a bug or just autobeam being its usual steaming pile of crap. For the next release, we really need to make it a priority to gut that shit once and for all and write something that works. Ctrl+B/U works
Dialog looks OK. I tested 4 in the time of 3 quarter notes with existing notes, and it looks OK, but I don't have my speakers turned on or have much of an intrinsic understanding of tuplets anyway. I've seen a lot of tuplet bugs on the main SF tracker, and don't expect we've fixed any of them, and I suspect that both Tuplet and Triplet applied to existing notes are dubious, but I don't think any of that is new, so it passes this inspection.
All checks out, along with ) and Ctrl+) hotkeys. Slurs seem to cause odd problems with sweep selections, but the effect is hard to pin down.
All checked out, including ~ hotkey
Getting the “can't create identical overlapping indications” error box in response to trying to replace a crescendo with a decrescendo is a bit confusing.
Passes very cursory inspection. Vertical layout has problems. I just threw one staff well into another staff's space by adding a 15vb.
This one works fine
Enter a half note a quarter note away from a barline, yielding a half note and a red barline. Use this function to split it across the barline into two quarter notes tied across. Works.
I've always despised this function. I have no idea if it's working worse than normal, but it's doing something at least.
For testing, I drew a simple segment in the matrix (works with raw duration) and then used the shift key to knock the timings and durations out of whack. Opened in notation, the segment had a lot of weird 64th rests. Various options had various tangible effects. Nothing quit cleaned up the example mess, but that's not an indication of breakage.
For testing, I took the notation view of the Quantize test segment alongside the matrix view of that same segment. Upon applying that, the contents of the matrix view (works with raw duration) shifted visibly. Looks like it's working to me.
Constructed a similar example to the one above. Applied some quantization, and then tried to remove it. It did something is as deeply as I'll explore this for this feature review.
Threw some dynamics and hairpins around, then looked at velocity in the matrix (since I can't see velocity in this context in notation). I can't say if the interpretations worked quite right, but this was a very messy test situation, and it did something at least. Dialog was fine.
Halve and Double work fine. Stretch or Squash is something I've never actually used. Dialog looks OK. Results were surprising, but I have no baseline. It does something at least, and is likely as good and as buggy as ever.
Works. Up / Down arrow keys also work.
Works. Ctrl+Up / Down also work.
You can merrily Ctrl+Up until the notes from staff 3 are above staff 1. We've always had vertical layout issues with things like this (using the same amount of space for everything, even when more or less space is indicated) but this seems worse than normal. (Although it actually isn't. I just tested and reproduced this with Classic. About 24 ledger lines or so. Nice. But not a regression, so I'm removing the bug flag and leaving this as an observation.
I have no clue what these do, but they do something.
(From here down the order is flipped bottom to top.)
Jog Left / Jog Right
Need to implement option 4 in off-list discussion. One property/pitch/controller n ruler per view. Ruler tracks active segment and updates its contents to reflect
Works swimmingly now that I've fixed the dialog.
Works generally with the caveat that selections between the edit view and any associated ruler are buggy. See...
I keep feeling like there's something still vaguely quirky about all of this, but I can't pin anything down at the moment.
I reviewed the behavior not long back and fixed the text background, resizing problem,and so on. I have not done a fresh review, but at a glance it seems to be in the state I left it in, and well-behaved enough for release.
This all seems to be working, but I did not test methodically this time around, and breakage may have crept in somehow since we were hammering on all of this a few months ago. Probably hasn't, but I'm admitting the lack of testing for the record.
I did not test every combination, but I've been through this pretty heavily while working out the problem with C and B and so on in the Romance language translations. That all works now, and I think it's all in good shape, but I must admit I did not test this thoroughly.
Observation: We really need to make up some keyboard maps à la Open Octave Midi and publish them somewhere. Even link to them from inside the GUI by way of web links or something. Doesn't necessarily have to be done to release, but it would be very nice of us.
Observation: I tried do-re-mi keys and toggling H to enter chords followed by single notes followed by chords, and it all worked well.
Even works in combination with grace mode. Good stuff. There are assorted well-documented age-old problems with triplet spanners and whatnot, but none of that impacts release.
Seems to be working as well as always. FIXED the stupid grace note scaling problem (more or less, I hope)
It took a bit of jiggling to get this to work, but I think it's OK now that I've saved a working MIDI input setup in my default studio. In spite of getting rid of the stupid “current? Y/N” problem (the answer is YES DAMMIT now) we still don't have working MIDI recording out of the box on my home setup, because there's only one input device, and it's not connected to anything at all out of the box.
This is one thing where I might consider the opposite approach from the one I've been going on about on the playback side. There's not much harm I can think of creating devices for every available RECORD device, and hooking them up out of the box. (Whereas doing this on the PLAYBACK side is completely EVIL.)
(Note, I did this in bottom to top order originally, and I've grown bored with rearranging it back into top to bottom order.)
Seems to work well enough, with some caveats I'm going to ignore until Abraham Darby unless they prove immensely serious
The Track Header blue outline moves responsively as I would expect.
BUG The playback/insert cursor stays anchored to its original spot BUG I had a whole staff selected. When I moved up and down from there, the selection remained
Entering with do-re-mi keys follows the blue track header outline. After entering the first note in this fashion, the two bugs above correct themselves.
If a loop is set manually, this does work.
Obeys “have selection” action state.
Dubious icons. I'm not sure about these, but I think I like them better than I did the last time I looked at them, so maybe we should leave well enough alone there. (Or remove this feature if it won't be made to work again.)
I won't go through these one by one. They seem to be working.
Choosing chords and inserting them into notation works generally.
BUG Something is wrong with the highlight colors. The highlighted fretboard has a different background from all the other highlights. May be a style problem. [Fsck it. Not even moving this to the Abraham Darby page.]
Seems to work and look rather better than it used to.
BUG From fret 5 onward, the little blue hover cursor does not show up, and it isn't apparent you can click to make a dot appear. [Fsck it. Not even moving this to the Abraham Darby page.]
Not very apparent which ones can be deleted and which ones can't, but the “Delete” button does cycle enabled states appropriately, and this does seem to work generally. General improvement might be indicated long after more pressing matters are dealt with.