updateChatSettings read the chat's contribution and unread tag before the
change from the listed chat after apiSetSettings returned. An item event
processed during the call can already carry the new settings (addChatItem
-> updateChatInfo), so before equalled after: the count and the unread tags
kept the chat's old contribution. Before is now the listed chat's stats
with the listed chat's settings from before the call.
Opening a chat replaces its chatStats with the server's on Android and
desktop. When the chat was already listed and the server reports fewer
unread items than the list had (e.g. items expired or were read on
another path), the count now drops by the chat's contribution
difference, like every other stats writer.
A content-filtered initial load of a group returns zero stats when no
unread item matches the filter (otherwise the full counts), so a load
with a content filter doesn't move the count (statsFiltered). A search
never reaches this branch: the core rejects initial pagination with a
search.
It only moves down. A chat from outside the 5000-chat window can be
listed with empty stats before it is opened, while the seed already
counted its unread items; moving up on its first open would count them
twice. Chats the load adds come with the server's stats, which the seed
already counted, so the count stays as it is for them. Upward drift
corrects itself at the next reseed: resume on Android and iOS, a profile
switch, opening the profile picker, or changing the chat item TTL.
If a new item arrives between the load's query and its stats
replacement, the downward move can drop the count below the database
until the next reseed.
iOS keeps the listed chat's stats on load, so there is nothing to move there.
The count is seeded by getUsersInfo, which counts RcvNew items where ntfsEnabled
holds - every item for an unmuted chat, only mentions for a mentions-only one,
none for a muted one. The deltas that maintain it afterwards did not follow that
rule for the active profile: addChatItem incremented for every item regardless
of the chat's notification setting; on Android and desktop the mark-read helpers
subtracted the full unread count, while on iOS Mark read and Clear followed the
mentions-only setting but subtracted the full count for muted chats, and
per-item reads forwarded the raw delta; removeChat left a deleted chat's unread
behind; and changing the setting moved a chat between rules without moving the
count.
Chat.userUnreadCount names what a chat contributes, and it is not a new rule:
totalUnreadCountForAllUsers already computed exactly this inline for the app
badge, and now calls it. The sites that change a chat's stats, its settings or
its presence report the contribution before and after, and the count moves by
the difference - apart from the exceptions noted at the end.
For that to hold the stats it reads have to be maintained, so on Android and
desktop decreaseCounterInPrimaryContext now decrements unreadMentions alongside
unreadCount when the item was a mention. Without it a mentions-only chat's
contribution never moved on any of that function's three paths - a deleted item,
a moderated member item, or an item going from New to Read. iOS already passed
the mention delta on its deletion paths.
This is visible beyond the profile picker, which badges only other profiles: the
desktop tray dot sums the count across profiles including the active one, and on
iOS these same deltas drive the app icon badge. So on iOS, muting a chat with
unread now drops the badge by its contribution, deleting a chat drops it, and a
message in a muted chat no longer raises it - each of which now agrees with
totalUnreadCountForAllUsers, which the badge is also set from.
addChat is deliberately not the mirror of removeChat: adding a chat means it is
being loaded into view, and its items were already counted by the seed, so
counting them again on ChatItemsLoader's addChat would double them. The one
place that removes and re-adds a chat while keeping its items - deleting a
contact in "keep conversation" mode on iOS - now carries its stats across and
restores the contribution explicitly, since the core keeps both the contact row
and its items there and the seed still counts them.
Wherever the chat list is re-read wholesale the counts are re-read with it: both
resume paths, and setting a chat item TTL, which deletes the active profile's
items without per-item events and so leaves its count stale. They are assigned
before the chats, which matters on iOS because updateChats recomputes the app
badge from users there. Reading users is best effort on both
platforms - losing them must not cost us the chats already fetched.
This does not make the count exact. getUsersInfo has no group_scope_tag filter
where the query behind chatStats.unreadCount does, so the seed counts
support-scope items that no client-side rule can, and the count steps up by the
outstanding support unread on each reseed. replaceChat also installs a whole
chatStats without moving the count: on Android and desktop from a snapshot taken
before the API call, in ChatListNavLinkView's markChatRead and markChatUnread
and ChatView's markUnreadChatAsRead, and on iOS from the server in the
receivedContactRequest handler in SimpleXAPI. Settings changes that arrive in
events through updateChatInfo do not move it either.
* desktop: fix crash in SimpleX address screen when there is no profile
* desktop: close SimpleX address screen when there is no profile
* desktop: rerun address screen self-close when its modal is reused or the stack changes
* desktop: close SimpleX address screen only when it is the top modal
* plans: correct address no-profile crash triggers and fix description
* plans: correct address no-profile crash fix description and verification
* plans: tighten address no-profile crash wording
* plans: add self-destruct trigger to address no-profile crash
* plans: qualify self-destruct trigger on desktop
* plans: drop desktop-only title from address no-profile crash
* plans: no connected mobile, not no mobile
* spec, product: document isLastModal and address screen without profile
* plans, spec, product: plan sections, address row spec, GAP-08 and GAP-09
* spec, product, plans: fix GAP-08/09 wording, table placement, ModalManager line
* spec, product: GAP-08/09 accuracy, gap count, clickable ModalManager location
* product, plans: GAP-09 at the level the code supports; qualify picker auto-open
* spec, product, plans: fix GAP-09 locations, stale line links, wording
* spec, product, plans: update plain-text line numbers; GAP-09 read wording
* plans: name the picker auto-open condition
* spec, plans: wording; security note
* plans: self-destruct did not crash on Android master; not a trigger
* plans: self-destruct path possible but unconfirmed; Android recovers silently
* spec, product, plans: drop GAP-09 and unrelated line refreshes
* plan: Android self-destruct path confirmed in logcat
* desktop: close Chat preferences screen when there is no profile, as the address screen does
* ui: show unicode characters in address as unicode markers
* defence in depth for links in messages (they are non-clickable from parser)
---------
Current code does not have this problem, as markdown parser rejects such links, and they are not clickable - it is needed for new markdown file viewer - this way unicode links in markdown show with unicode sequences instead of look-alike letters.
* rfc: invite member of one group to another group
* update
* update
* implement
* version
* update
* do not show error if contact was removed when joining group
---------
Co-authored-by: Evgeny @ SimpleX Chat <259188159+evgeny-simplex@users.noreply.github.com>
* android: recognise passphrase fields as passwords for password managers
Passphrase fields used a hand-rolled VisualTransformation lambda, so Compose never
marked them as password fields: the autofill structure carried no hints and no
password inputType, and password managers could not identify them. The raw
passphrase was also exposed through the CopyText/ACTION_COPY accessibility action.
PassphraseField now declares an autofill content type and uses
PasswordVisualTransformation, which sets the password semantics Compose derives
that flag from.
Fields that create a passphrase declare NewPassword. Where no field would carry
plain password (first-time setup, hidden profile password) they declare both, as
managers do not recognise a form whose every field is newPassword. On the change
form the current-passphrase field supplies password, so the new fields declare
only newPassword - adding password there leaves the manager unable to tell which
field holds the new value and suppresses the save prompt.
* android: do not offer to save hidden profile passwords
Hidden profile password fields no longer declare an autofill content type.
Compose commits the autofill session, which shows the password manager's
save prompt, only after the last field with a content type leaves the
screen, so these fields never trigger it. A saved entry would reveal that
the hidden profile exists.
Rewrite the plan to match the code.
* android, desktop: storage breakdown in developer options
Port of the iOS Storage screen (#5529): lists each top-level entry of the app's data, config and temp folders with its recursive size, so it is possible to see what takes space.
* android, desktop: simplify storage view
Use DefaultProgressView, drop the bold total and redundant absoluteFile, and do not follow top-level symlinks (matches iOS; avoids counting the Android lib link).
* android, desktop: storage view: do not crash on names the JVM cannot convert to Path
Under a non-UTF-8 locale on Linux, File.toPath() throws InvalidPathException for a non-ASCII top-level name (e.g. a received file decrypted into tmpDir); log it and show 0 bytes. Also correct plan wording.
* android, desktop: storage view: fold path conversion into treeSize, use produceState
Also correct plan wording (du per-directory overhead, unlistable roots, Windows junctions).
* android, desktop: storage view: open as card screen like other settings screens
* android, desktop: storage view: list top-level entries as Path to keep their exact names
File.listFiles + toPath loses names that are not valid in the JVM file-name encoding (non-ASCII under a C locale, invalid UTF-8), so those entries showed 0 bytes. Files.newDirectoryStream keeps the raw bytes.
* android, desktop: storage view: normalize root paths before dropping nested roots
A root spelled with .. (e.g. XDG_DATA_HOME=/x/data/../tmp) compared as nested in another root and its size was shown nowhere.
* android, desktop: document storage view in spec and product docs
* android, desktop: storage view: single catch when listing a root
* android, desktop: do not log file paths in storage view
The walk's IOException messages contain the full path, which in tmpDir can be a received file's original name.
---------
Co-authored-by: Evgeny Poberezkin <evgeny@poberezkin.com>
The core allocates every chat_* response with malloc and passes ownership to
the caller, as iOS does in API.swift. Both JNI bindings copied the bytes into a
Java string and dropped the pointer, so every command, event and file read
leaked its whole response into the C allocator - invisible to both the JVM and
the Haskell heap profiler.
Desktop frees in decode_to_utf8_string, the funnel every entry point already
goes through. Android has no such funnel because NewStringUTF is called inline,
so decode_and_free is added and the call sites routed through it. chatReadFile
frees directly on both.
Co-authored-by: Evgeny Poberezkin <evgeny@poberezkin.com>
* tests: close the walk stream in withTempDir
* desktop: receive app links from args and macOS
* desktop: forward an app link to the running app
* desktop: initialize the desktop platform lazily
* tests: share the temp directory helper
* desktop: register simplexchat scheme in packages
* desktop: register simplexchat scheme at runtime
* ui: end the badge page with a link on desktop
* desktop: raise the window over other apps
* docs: describe desktop deep links in the spec
* docs: plan desktop deep links
* docs: report on desktop deep links
* desktop: open simplex connection links
* docs: describe simplex links on desktop
* desktop: keep a link from opening twice
* desktop: retry taking a signal held open
* desktop: keep the user's AppImage link handler
* desktop: register Windows links through Capabilities
* desktop: add %u to the deb entry only once
* docs: describe the review fixes
* ui: open the badge link screen on the left
* ui: import moko accessors in the badge screens
---------
Co-authored-by: spaced4ndy <8711996+spaced4ndy@users.noreply.github.com>
* core: group fixes/improvements
* Large messages dropped by relays and hosts (the same 15602 limit was applied, without accounting for forwarding header)
* Members below v18 did not receive forwarded messages or history in p2p groups.
* Join confirmations near the size limit could fail.
* SMP servers are limited to 2 host names each.
* simplify
* remove legacy forwarding and history support
* only restrict p2p groups
---------
Co-authored-by: Evgeny @ SimpleX Chat <259188159+evgeny-simplex@users.noreply.github.com>
* android: create new keystore keys in StrongBox
* android: show keystore key storage level
* ui: save keystore key before marking it stored
* android: show keystore storage level in the footer text instead of a separate row
* android: return keystore storage label, drop enum
---------
Co-authored-by: Narasimha-sc <166327228+Narasimha-sc@users.noreply.github.com>
* core, ui: remove database backups after key changes
* ui: delete database backups on chat start without delaying encryption marker reset
Keeping encryptionStartedAt set until chat start made DatabaseErrorView show
"The attempt to change database passphrase was not completed" with a Restore
button after every successful passphrase change when the passphrase is not
stored. Restore the marker handling from master and delete backups
unconditionally after chat starts.
* another flag
---------
Co-authored-by: Evgeny @ SimpleX Chat <259188159+evgeny-simplex@users.noreply.github.com>
Co-authored-by: Narasimha-sc <166327228+Narasimha-sc@users.noreply.github.com>
"Blur media" hid images and videos in a chat but showed the same media
sharp as the thumbnail of a quoted message and as the last message's
image, video or link preview in the chat list.
Small views are now blurred with the same reveal as chat media. They are
resampled relative to the width they are drawn at, so a thumbnail is
blurred as much on screen as media in a chat, and drawn as the centre
crop the thumbnail shows. In-chat media keeps the same blur.
Blurring the chat list also needs: each last message gets its own blur
state, loaded chat-list videos use it and hide their full-screen play
button while blurred, a right-click on a thumbnail no longer leaves it
revealed on desktop, and leaving a chat mid-scroll no longer undoes
reveals in the list on Android.
Co-authored-by: Narasimha-sc <166327228+Narasimha-sc@users.noreply.github.com>
* ui: blur media by resampling the preview, not a per-frame effect
Modifier.blur put a BlurEffect on a display-sized graphics layer, so the
Gaussian was re-evaluated on every frame the media was drawn. It also
requires RenderEffect, which Android applies only from API 31, so below
Android 12 media was drawn unblurred while the setting read as on.
A blur and a downscale discard the same thing - detail finer than their
radius - so the preview is now resampled to about one pixel per radius and
stretched back. That runs once when the item composes, needs no
RenderEffect, and removes the detail irreversibly rather than convolving
it. The first resampling step bounds both sides, so no image, however
shaped, can produce a large intermediate.
While the blur hides the media the file is also left unread, so an image
scrolled past is no longer read, decoded at its full size and kept in the
image cache only to be hidden again. blurHidesMedia() is the single
definition that both the drawing and the loading decision use, so they
cannot disagree about whether the media is on screen.
* plans: blur media by resampling the preview
* ui: shorten comments in media blur
* ui: smooth the media blur and match its strength to the old one
Stretching the resampled preview straight to the screen showed its pixel grid. Double it back
up until its longest side is at least half the reference width before drawing, so the last
stretch is short and the tents compose into a bell. The reference width moves from 360 to 400:
fitted against Gaussians, 360 blurred 5-75% more than Modifier.blur did, 400 is within 10% at
every setting.
(cherry picked from commit c0344a9a61)
Narasimha-sc