"Backup" does not fully back up all presets ("sweep desynced, aborting")

[rtsai]

Test configuration:

• Hardware: HX Stomp XL firmware 3.80.0
• fretwire: commit c99c837

While testing #2, I clicked "Backup", and encountered some instances of

backup: backup file: device reports preset 1 while backing up slot 2 - sweep desynced, aborting

The progress through the backup seems to be affected by which preset is currently selected in the UI.

It is not immediately clear whether or not this is directly related to #2 (I had not tried "Backup" before).

A propos, I suggest renaming the UI buttons to be more consistent with the HX Edit UI (e.g., "Backup" seems to be more of a "set list export" rather than a "Device backup").

[john-baxter-dev]

Thanks for reporting this, and for including the exact message — the off-by-one and the "depends on which preset is selected" observation together pointed straight at it.

After a preset change the device answers the identity query with the previous preset's address for a few hundred milliseconds. That's known, and read_preset already waits it out, but the backup sweep read the identity once and believed it — so it saw slot N−1, decided the sweep had desynced, and aborted a sweep that was fine. The start-of-sweep read was in the same window, which is why it tracked the selected preset: the button is usually pressed right after clicking one.

Both now go through a single confirmed read that waits for the address actually asked for. Reproduced live on an HX Stomp: a full 126-slot sweep, with the log showing the lag happen and be absorbed rather than abort. Fixed in 4c97a38.

Your naming point was right, and it was covering a real bug. The sweep always walked bank 0 regardless of which setlist was on screen — harmless on a single-setlist device, wrong on a Helix. It's now "Export presets" (CLI export-setlist; backup still works as an alias), exports the setlist you're viewing, offers all setlists with a cancel, and says plainly that it holds presets and not globals or IRs. A genuine device backup is on the roadmap behind those two pieces; until it exists, nothing gets called one. That's e23d003.

Also a quick question: How does the pedal's screen number presets — 01A/01B/01C/02A etc like the HX Stomp? The editor now labels presets the way the panel does, but only for the HX Stomp — the XL keeps plain slot numbers until we know its banking, rather than guess wrong.

[rtsai]

How does the pedal's screen number presets — 01A/01B/01C/02A etc like the HX Stomp?

On HX Stomp XL the presets are actually in groups of 4, A/B/C/D, so 000-127 or 01A-32D, depending on Global Setting preference. I have read the HX Stomp has 3 snapshots pre preset (HX Stomp XL supports 4 snapshots per preset), so maybe the HX Stomp grouping presets into groups of 3 (A,B,C) is related to that as well?

Maybe this image will explain what I mean:

• Potential image source, but line6.com seems to be down right now: https://line6.com/support/topic/66676-hx-stomp-xl-assigning-fs1-bank-down-to-a-snapshot/
• Cached image from Google search: https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcR5myOyDLWCnZYRHGK5ZgppbvIWRdKMh21vGGF5NK2560rOLeV15gXyB_Q&s=10

I couldn't find any similar images for HX Stomp (non-XL) so I don't know what the HX Stomp looks like.

[john-baxter-dev]

Oh awesome, much appreciated!! I will implement that... I didn't realize you could switch to 0-127 numbering as well, I actually prefer that. I'll make it configurable in Fretwire as well!

[rtsai]

I'm not sure this is fully addressed, as not least not in an end-to-end fashion.

With commit c936aec, it got up to preset "23C" (091?) then stopped:

[john-baxter-dev]

Hm okay thanks for letting me know. Let me take another look!

[rtsai]

There is a transient issue, possibly still related to #2

• The UI had some missing rows (approximately 009-015 or so), and 008 had a blank item in the preset list, although the main panel showed the correct preset name
• The console logs showed:

2026-08-21T03:11:02.826838Z  INFO fretwire_core::session: read-info reply (current preset identity) preset_info=Some(PresetInfo { bank: 0, index: 7, name: "KravitzSoup", snapshot: Some(0) })
2026-08-21T03:11:02.848320Z  INFO fretwire_core::session: stream-start reply (chunk #0) arg=6179 body=256
2026-08-21T03:11:02.856435Z  INFO fretwire_core::session: reassembled preset stream bytes=3419 declared=3419
2026-08-21T03:11:02.894247Z  INFO fretwire_core::session: read-open reply arg=6234 body=76
2026-08-21T03:11:03.014717Z  INFO fretwire_core::session: read-info reply (current preset identity) preset_info=Some(PresetInfo { bank: 0, index: 7, name: "KravitzSoup", snapshot: Some(0) })
2026-08-21T03:11:03.109548Z  INFO fretwire_core::session: stream-start reply (chunk #0) arg=6288 body=256
2026-08-21T03:11:03.117360Z  INFO fretwire_core::session: reassembled preset stream bytes=3366 declared=3366
2026-08-21T03:11:03.121548Z  WARN fretwire_core::session: preset identity moved across the stream read — blob provenance is ambiguous before=(0, 7, "KravitzSoup") after=(0, 8, "EverlongPlexi")
2026-08-21T03:11:03.381652Z  INFO fretwire_core::session: read-open reply arg=6322 body=76
2026-08-21T03:11:03.383684Z  INFO fretwire_core::session: read-info reply (current preset identity) preset_info=Some(PresetInfo { bank: 0, index: 8, name: "EverlongPlexi", snapshot: Some(0) })
2026-08-21T03:11:03.404620Z  INFO fretwire_core::session: stream-start reply (chunk #0) arg=6376 body=256
2026-08-21T03:11:03.412762Z  INFO fretwire_core::session: reassembled preset stream bytes=3366 declared=3366
2026-08-21T03:11:03.450412Z  INFO fretwire_core::session: read-open reply arg=6431 body=76
2026-08-21T03:11:03.569688Z  INFO fretwire_core::session: read-info reply (current preset identity) preset_info=Some(PresetInfo { bank: 0, index: 8, name: "EverlongPlexi", snapshot: Some(0) })
2026-08-21T03:11:03.591679Z  INFO fretwire_core::session: stream-start reply (chunk #0) arg=6485 body=256
2026-08-21T03:11:03.599748Z  INFO fretwire_core::session: reassembled preset stream bytes=3371 declared=3371
2026-08-21T03:11:03.603420Z  WARN fretwire_core::session: preset identity moved across the stream read — blob provenance is ambiguous before=(0, 8, "EverlongPlexi") after=(0, 9, "EverlongPlacater")
2026-08-21T03:11:03.846375Z  INFO fretwire_core::session: read-open reply arg=6519 body=76
2026-08-21T03:11:03.848889Z  INFO fretwire_core::session: read-info reply (current preset identity) preset_info=Some(PresetInfo { bank: 0, index: 9, name: "EverlongPlacater", snapshot: Some(0) })
2026-08-21T03:11:03.869621Z  INFO fretwire_core::session: stream-start reply (chunk #0) arg=6573 body=256
2026-08-21T03:11:03.877788Z  INFO fretwire_core::session: reassembled preset stream bytes=3371 declared=3371
2026-08-21T03:11:03.923099Z  INFO fretwire_core::session: read-open reply arg=6628 body=76
2026-08-21T03:11:04.028053Z  INFO fretwire_core::session: read-info reply (current preset identity) preset_info=Some(PresetInfo { bank: 0, index: 9, name: "EverlongPlacater", snapshot: Some(0) })

However, when I disconnected and reconnected, both issues went away:

• UI fully rendered all preset names
• All presets successfully exported
• The logs still show the messages similar to preset identity moved across the stream read

[john-baxter-dev]

Thanks for pushing back on this — you were right that it wasn't fixed end to end.

The guard that produced that message was doing its job: that blob really was corrupt. What was missing was recovery. The GUI's read has always retried a bad stream, but the export used a different read path that propagated the error straight out — so browsing survived a splice and a sweep died on the first one. Both retry now.

The missing sidebar rows look like the same transient in the browse listing rather than anything to do with #2. That path had no retry at all and only checked for a short read, and a listing that runs long still parses — just into the wrong rows. Same guard and same retry now.

I'm leaving this open rather than closing it, because I can't confirm the fix from here — a 126-preset export on my Stomp ran clean, but nothing spliced, so the retry never actually fired. If you re-run the export and it completes, that's the confirmation. If it stops again, the log will now tell us which slot and how many retries it burned before giving up.

Also: the XL now banks by four (01A–32D), thanks for that. You were right that the display form is a Global Setting — I tested it on a Stomp and it never reaches the host at all (byte-identical listing and preset stream in both settings), so there's now a toggle in the ⋯ menu to match whichever your pedal is on.

On the snapshot theory: the Helix Floor has eight snapshots but almost certainly still four presets per bank, so the two probably aren't linked — though nobody's read us a Floor's screen yet, so I can't say for certain.

Side effect of chasing this: preset switching is about 35% faster and does half the USB traffic.

Let me know if this resolves it!

[rtsai]

I tested against commit 5dee6bc and things seem to work, and setlist export does seem snappier, although I haven't measured it. I do note that I no longer see any logs about preset identity moved across the stream read — blob provenance is ambiguous

Are you able to run HX Edit? I currently use it inside of a Windows virtual machine inside of VirtualBox, solely to back up my device. (I do all my editing on-device, just because Windows-inside-Virtualbox is so painful.)

The reason I ask about your use (or non-use) of HX Edit is to understand whether the internal data modeling is kind of being completely reverse-engineered bottoms-up-from-scratch from on-device workflows only, or if it is somehow being informed "top-down" from the HX Edit UI.

FYI, the fretwire button labeled as "Export presets to file" seems to be the equivalent of what the HX Edit UI calls "Setlist export" (but I don't know what the HX Edit UI does on a multi-setlist device like Helix, or maybe they have a different "Helix Edit" desktop program).

Both fretwire and HX Edit end up exporting a single JSON blob of all the presets, but no IRs, Favorites, Global EQ, Global settings, user defaults, user preferences, etc.

The HX Edit "device backup" is does not appear in the visible GUI and is hidden inside a menu bar option.

[rtsai]

(I consider this issue fixed and may be closed.)