 I've just completed ADR-0035 on remembered alterations: the system determining when accidentals print in music notation. The client's first window won't open for yet a few weeks, so there's no technical pressure for this work. But conceptually, it felt necessary. Pitch representation, time signatures, key signatures: all complete. Before proceeding to visualization, I needed to see these systems work together. I needed conceptual closure. The result surprised me. Not because it works (of course it works), but because of how naturally the pieces fit. The Actual ProblemAccidental rendering isn't 'print a sharp when the note differs from the key signature'. Accidentals have memory. Once F♯ appears in a measure, subsequent F notes remember that sharp until the barline. This memory flows left-to-right through musical time, independent of visual layout. The complexity emerges in multi-staff instruments. Piano, harp, organ, even choirs: all staves share accidental memory. A sharp in the treble affects naturals in the bass. Multiple voices complicate further; temporal sequence follows rhythmic position, not staff order or voice structure. You need a model where musical time is explicit and independent of how things appear on the page. This is precisely the problem that four separate architectural decisions, made for entirely different reasons, were built to solve. What Was Already There Pitch representation (ADR-0026): Letter names, octaves, accidental keywords. Built for memory efficiency and canonical identity. No thought given to accidental rendering at the time. Time signatures (ADR-0033): Metric positions as rationals (1/4, 3/8, 5/16) providing exact temporal coordinates for every event. Built to handle irrational and additive meters correctly. Again, no consideration of accidental logic. Key signatures (ADR-0034): Define which accidentals are 'normal' for a given context. Standard keys, keyless modes for keyless music, per-octave microtonal signatures for contemporary notation. Built as baseline, not as accidental memory. Timewalk (ADR-0014): Traverses music in strict temporal order. Built for efficient general-purpose traversal. The musical timeline made explicit as implementation detail. Four systems, four separate problems, no coordination between them. The Single Comparison RuleWith those foundations in place, remembered alterations became straightforward. The algorithm is a wave pattern flowing left to right:
Decision rule: A note requires a printed accidental when its accidental differs from the current remembered state for that letter and octave. That's it. This single rule handles both required accidentals (contradicting key signature) and courtesy accidentals (restating after alteration). No special cases. No heuristics. The data structure stores only deviations from baseline; eight entries typical versus seventy for full representation. Grace notes participate fully in the memory system. Tied notes bypass conditionally with position-based French tie distinction. Simultaneities detect conflicts where the same letter has different accidentals at the same moment. Cross-octave courtesy accidentals work through letter-first grouping that makes the lookup O(m) where m equals octaves for that letter, not O(k) for all octaves globally. All of this through the same mechanism. The complexity isn't in the algorithm; it's in having the right foundations.  The Mental ShiftTimewalk changed how I think about notation problems. Instead of 'iterate over notes and check context', the remembered alterations pipeline is stream transformation:  Each stage is pure transformation. Grace notes are repositioned to exact temporal locations based on tempo-dependent duration (85ms default, compressed on collision). Simultaneities group by rhythmic position. Conflicts detect same-letter different-accidentals at the same moment. The reducer threads remembered state through the stream, accumulating decisions. This pattern will handle dynamics, articulations, phrase marks: any notation requiring temporal context. The abstraction scales because the foundations were right to begin with. What This Actually RevealsThe algorithm produces semantic decisions about accidental requirements, independent of layout. These decisions are deterministic: same music, same decisions, always. They're layout-independent: work identically across all visual representations. They're configurable: house style settings modify behaviour without code changes. They're correct: edge cases handled through the same mechanism as common cases. But here's what matters: I didn't design these four systems to work together. I designed each to solve its own problem correctly. The alignment emerged because correct solutions in related domains tend to compose naturally. This is evidence. Either the domain model reflects musical reality accurately, or I've successfully imposed coherent structure on chaos. Those are different conclusions with different implications, but both suggest the foundations can carry what comes next. The Timing QuestionWhy now? Why remembered alterations before any visualisation exists? Because I needed to know whether the foundations were actually complete. Pitch, rhythm, key signatures, temporal traversal: individually validated, but never tested as a system. Remembered alterations forced that test. If they didn't align naturally, I'd have learned something important about what was missing. They aligned. Beautifully. With no compromises in existing architecture, no special cases, no friction. Four systems built independently composed into a solution more elegant than any I'd planned. That's conceptual closure. That's knowing the foundations are right before proceeding to build on them. That's why this work happened now, before technical necessity demanded it. The next phases (windowing, rendering, note entry) will build on these same primitives. But now I know they're sound. Not because I believe they should be, but because I've seen them work together when they had no reason to unless the underlying model was correct. The window opens soon. But first, this needed to be right. Now I know that when notes appear on screen, there will be no special cases left to fix, no rules-of-thumb to implement. Accidentals are now correct regardless of staff, voice, cross-staff, or cross-measure situation or house rule preferences. Es ist alles eins, as Marie sings in Wozzeck. But in a good way. 
5 Comments
 Key Signatures in Ooloi have been specified and implemented, and as a result we have a new Architectural Design Record, ADR-0034, on the Key Signature Architecture. Standard Key SignaturesIt covers standard key signatures – the usual and expected: major, minor and modal in all keys:
 Key signatures with double flats or sharps are also supported:  Ich fühle Luft von anderem PlanetenOoloi also supports keyless signatures, custom key signatures mixing flats and sharps, as well as microtonal key signatures.    Ooloi also handles key signatures where accidentals differ between octaves (for those moments when you feel your beard turning blue). PlaybackAll this works with playback, of course, and even the microtonal key signatures handle cautionary accidentals between voices and measures as expected, according to the engraving principle of 'remembered alterations'. A note on the ADR: It's not the UIThe ADR shows the internal architecture - what the software does behind the scenes. Users enter key signatures through standard dialogs: 'Eb major', 'D Dorian', etc. The technical infrastructure matters because it's what enables all those custom and microtonal features to work seamlessly alongside standard notation.
 Fractional numerators are now official. Ooloi’s time-signature system (ADR-0033) supports everything from common time to irrational and additive meters, and now fractional counts like “2½ / 4”. Halves only, for now – enough to cover Grainger and Chávez without venturing into Boulezian fantasy. Full ADR: ADR-0033-Time-Signature-Architecture  Henri Dutilleux: Choral et variations  Alfred Hitchcock's Vertigo opens with spirals. They appear in the credits, in Carlotta's hair, in the cross-section of the redwood, in the geometry of the bell tower staircase. Spirals aren't circles; they're recursive ascents, passing the same point at different heights. Each loop looks familiar yet transformed. In the film's second half, Scottie Ferguson attempts the impossible: reconstructing a lost woman. He obsesses over every detail – the hair, the suit, the mannerisms – trying to make Judy become Madeleine. The reconstruction is meticulous, desperate, ultimately doomed. Madeleine was never real. She was always performance, designed to deceive. I've been obsessed with this film for decades. Vertigo asks the question I cannot stop asking: what constitutes identity across time? Can something be reconstructed, or does reconstruction merely create elaborate performance? When does architectural continuity become delusion? Twenty-five years ago, I built notation software called Igor Engraver. Then the world moved on, and the code became archaeologically interesting but technologically obsolete. Common Lisp running on pre-OS X Macintosh, rendering via QuickDraw GX, single-threaded on single-core CPUs. Now I'm rebuilding it as Ooloi – new language, new rendering engine, new concurrency model, new architecture. So I have to ask: am I just Scottie in Vertigo? Reconstructing a past that never really existed?  The Fundamental Difference Madeleine never existed. Igor Engraver did. Scottie reconstructs a performance designed to deceive him. I'm reconstructing software that worked perfectly well with the technology it had. Igor Engraver didn't die from technical inadequacy. It died from narcissistic neglect by guys with ponytails who couldn't see past their own reflections. The architecture was sound. QuickDraw GX handled what we asked of it. Common Lisp did exactly what we needed. The software worked. What failed was human commitment, not technical capability. Ooloi takes the same architectural vision and asks: what becomes possible when you're free of narcissistic partners and have access to genuinely better tools? The answer: not survival, but fulfilment of what was always possible. For, which is important: unlike Scottie, I'm not reconstructing the same woman. That's another significant difference. Igor's strengths lay in its ease of use and its user interface (and its playback, but that's another matter). Ooloi must go much further than Igor ever did in terms of typographical beauty, and it must do so from the first release. Herrmann's Transformed Tristan Bernard Herrmann's score for Vertigo doesn't just accompany the psychological state; it induces it. Those falling, spiralling arpeggios. Obsessive motifs that keep returning. Chromatic harmonies that never resolve. Lush orchestration that simultaneously feels unstable. The music is Wagner's Tristan und Isolde transformed – the same chromatic restlessness, the same endless longing, the same refusal to resolve. Wagner created music about impossible love that could never find harmonic rest. Herrmann took that DNA and made it cinematic, made it about obsession itself rather than the object of obsession. Years ago I quoted four measures from Herrmann's Vertigo in my opera The Maids – using his music about false love to underscore Genet's performed identities. Servants performing their mistress, performing revolution, performing each other. Identity as pure theatrical construction with no stable referent. The circularity is almost too perfect: the composer of Vertigo scoring the film about vertigo, whose music I once quoted to mark falseness, now frames my own question – whether I'm caught in the same spiral of reconstruction. But notation software can prove itself. Ooloi either handles Strauss's Elektra without freezing or it doesn't. It either supports real-time collaboration or it doesn't. Unlike Scottie's obsession, mine has to compile. What Constitutes Identity?This has always been the question. Vertigo asks it. The Maids asks it. Every recursive project asks it. Is Judy really Judy if she becomes Madeleine? Was Madeleine ever real? Does performed identity become real identity through sufficient commitment? Is Ooloi Igor Engraver, or am I performing the identity of my younger self while building something entirely new? Most programmers rebuild things pragmatically – new requirements, new tools, new project. But I'm asking whether I'm still the same architect when the materials have completely changed. Whether continuity of vision across technological discontinuity constitutes identity or elaborate self-deception. The technical choices answer the question. I preserved the idea: architectural vision, musical semantics, core abstractions, the fundamental insight that notation software requires deeper engineering than the industry has provided. But I let the implementation transform completely: Clojure instead of Common Lisp, STM instead of locks, GPU rendering instead of QuickDraw, immutable data structures, client/server architecture. That's what genuine identity actually is: essential persistence through necessary transformation.  The Ooloi Metaphor Becomes Literal Octavia Butler's ooloi are alien beings who perform genetic transformation – taking DNA from incompatible species and making them compatible. They preserve what's essential while enabling survival in new environments. They don't fake compatibility; they genuinely reconstruct at the genetic level. That's what I'm doing with software architecture. The DNA persists: core musical semantics, hierarchical design, fluid user interface that doesn't get in the way. Add to that massive parallelism, a pervasive plugin philosophy, and the realisation that collaborative capabilities are a by-product of the architecture itself. The organism transforms to survive in 2025: different language, different rendering, different concurrency model. The Ooloi name wasn't whimsy. It was recognition that genuine transformation requires going deeper than surface reconstruction. The Spiral AscendsScottie's spiral destroyed him because the object was always false. The reconstruction was doomed because there was nothing real to reconstruct. My spiral can succeed because the architecture was always sound; what failed was human commitment, not technical possibility. Igor Engraver died from neglect, not inadequacy. The guys with ponytails moved on. I did too, from the circumstances. But not from the subject. Twenty-five years later, the obstacles aren't technological maturity – they're human. I can build this alone, without partners whose narcissism exceeds their commitment. I can use tools that are genuinely better (Skia, STM, cloud architecture) without needing them to justify the project's existence. The obsession is real. The vertigo is real – that uncanny feeling when you realise what was abandoned can be completed. The psychological pattern is acknowledged. But where Scottie's reconstruction existed purely in romantic space with no objective validation, mine has material proof: code that compiles, tests that pass, collaborative editing that actually works, Strauss's Elektra rendering in real time. Identity is architectural continuity, through material transformation and human liberation. Not performance. Not even evolution. Completion. The spiral passes the same point, but at a different height. This time, without the ponytails and the pop zombies.  Vertigo remains my favourite film. I understand Scottie's obsession. I recognise the danger. But I also know the difference between reconstructing what was never real and finishing what should never have been abandoned.
The partners are gone. So are the commercial pressures. The new architecture holds. Herrmann's spiral ends in unresolved ascent – motion without final cadence. Ooloi's must too.  You must remember this a bis is just a bis a sign is just a sign The fundamental things apply as time signatures go by Time signature support in Ooloi, from common to esoteric. Plain VanillaOoloi of course supports the usual:  ... and in 'the usual' I include things like time signatures spanning multiple staves in scores for readability purposes, etc. You know, what you'd expect from a modern, professional program. Intuitive EntrySpecifying a time signature is simple: you just write your signature as a string in the input dialog: "4/4", "C", "2/2", etc. As you will see, this paradigm holds for very complex or unusual meters, too. Additive MetersAdditive meters are of course also supported:  which simply is entered as "3+2+3/8".  which is entered as "2+2+2+3/8". You get the idea. You can do things like this too:  This would be input as "1/4+1/4+3/8+1/4". Use whitespace for clarity if you like: "1/4 + 1/4 + 3/8 + 1/4". You can combine this with multiple terms:  ... which simply is entered as "3+2/8 + 6/16". Orff, Fortuna!Carl Orff invented a notation showing beat duration as a note symbol rather than a number. It's occasionally useful, so Ooloi supports it: or They are entered as "3/8" and "2/4.", respectively (don't miss the full stop at the end of the second string). There's a checkbox to trigger this type of formatting – but as the second example contains a dotted value, Ooloi will detect the latter case automatically. Fractional MetersBoulez, Ferneyhough, Finnissy et al. Yes, Ooloi will support fractional meters such as 4/3 or 4/5. I personally would never use them, but then it's not up to me to pass judgement on what I think is musical over-engineering. That's what conservatories are for. Consequently, Ooloi supports the following:  |
AuthorPeter Bengtson – SearchArchives
January 2026
Categories
All
|
RSS Feed