Summary:
We need to remove quoted text completely from bodies in a composer.
Unfortunately, that makes it very difficult to determine how to put it
back in.
For now the scheme is to append the quoted text at the end. However, that
means that we need to only pull out quoted text at the end of a message.
Unfortunately there are lots of places that quoted text appears inline
with regular text.
Determining whether or not some content is at the "end" of a message
turned out to be non-trivial.
We now have a new `DOMUtils` that looks for empty areas at the end.
We also have a new quoted HTML parser that finds trailing quotes.
Fixes T2335
Test Plan: lots of new quoted text tests
Reviewers: bengotow
Reviewed By: bengotow
Maniphest Tasks: T2335
Differential Revision: https://phab.nylas.com/D1773
Summary:
- You cannot drag an attachmentComponent unless it represents a fully downloaded file on disk.
- You cannot drag and drop an item from a composer onto the same composer.
- Attachments on expanded messages start downloading as soon as it's viewed (necessary for drag and dropping attachments, also probably best. Previously was just images.)
- You can now cancel an attachment download. Was really broken.
Test Plan: Run many new specs!
Reviewers: evan
Reviewed By: evan
Maniphest Tasks: T1196
Differential Revision: https://phab.nylas.com/D1764
Summary: query trimming was broken due to searchquerycommited passing an object instead of a string
Test Plan: tested manually. Unit tests
Reviewers: bengotow
Reviewed By: bengotow
Differential Revision: https://phab.nylas.com/D1763
Summary: The diff adds very basic versioning to the database via sqlite's built-in `user_version`. If the version is bumped in DatabaseStore, it means that all existing data should be blown away and the user should have to refetch the entire cache. Critically, this does not log the user out.
Test Plan: Run no new tests :-(
Reviewers: evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D1760
Summary:
There are now two objects, Folders & Labels. These inherit from `Category`
(that's what Eben said they were using on the backend).
There are two separate tasks.
1. MoveToFolderTask
2. ApplyLabelsTask
It turns out that the semantics between the two are quite different.
The reverse operation for moving to a folder is a bit tricky.
As of 7-8-15, the Tasks are pretty much complete. I need to write tests
for them still and do some manual testing in the client.
Test Plan: Writing specs
Reviewers: bengotow
Reviewed By: bengotow
Differential Revision: https://phab.nylas.com/D1724
Summary:
The issue was that on every key stroke the whole composer, participants
and all, were getting re-rendered. According to React.perf, the
`TokenizingTextField`s were taking a very long time to render and never
changing. This was fixed by adding a simple `shouldComponentUpdate` check.
The composer also has several regions that only change when the `props`
do. These are now cached. The cache reset when the `props` do.
After all of that, rendering the whole composer still takes 20-40ms. If
you're tying in the composer very quickly, text entry can approach that
render time. This starts to stack multiple React rendering passes up and
bogs the whole system down.
Luckily, we can simply render the composer less frequently. Now, after
changes are persisted to the `DraftStoreProxy`, we simply debounce the
proxy `trigger`. The users don't see this because the native
`contenteditable` field will update immediately. When the debounced proxy
trigger fires, it will transparently update the view to the latest state.
Test Plan: edgehill --test
Reviewers: bengotow
Reviewed By: bengotow
Subscribers: mg
Differential Revision: https://phab.nylas.com/D1749
React-dnd is a nice library but it assumes it's going to be the only thing managing your drag and drop in the entire app. It also hides the underlying dataTransfer object so it's difficult to implement native drag and drop behaviors other than for Files and Urls, which they special-case.
Implementing it manually really isn't hard, and we can do things like attach a text/plain version of contact data to the contact drag.
Fixes T2054
Summary:
Cancelling a file upload properly makes the task return Task.Status.Finished (previously it was broken and returning Task.Status.Retry)
Dragging a directory now gives you a nice error message
Aborting an upload now looks for the task with the given ID, not a task with the given filepath since there could be identical uploads happening in other windows
Make it more explicit that the uploadId is not the task ID
Add a few tests
Test Plan: Run new tests
Reviewers: evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D1750
Summary: If less than 5 new unread notifications, then show notification for each new message. Each message is separated by 1 second
Test Plan: Added test to make sure Notification is called multiple times
Reviewers: bengotow
Reviewed By: bengotow
Maniphest Tasks: T1295
Differential Revision: https://phab.nylas.com/D1744
Summary:
fix(drafts): One failed promsie breaks ChangeSet.commit forever
Move the quoted text button so it's beneath the quoted text not at the bottom of the composer, also fixes problem with it overlapping content slightly
When the main window crashes, reload it immediately
Test Plan: Run tests
Reviewers: evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D1742
Summary:
consolidate all the "untitled" stuff into a convenience method on the File itself. Previously it'd show "Unnamed Attachment", download as "Untitled" and open as "<file.id>". Now it's consistent everywhere and chooses names based on the contenttype (Event.ics).
Rewrite CSS rules for uploads and attachments to be simpler
- remove container divs and classnames from things that have no CSS
- switch to using Flexbox so it's not necesary to have so many containers
- remove zIndex hacks, apply overflow rules to name div only, so long filenames don't make action button unclickable
- consolidate CSS classnames for uploads/attachments
-
Other style fixes
- cursor "default" instead of text insertion on image attachments
- cursor "default" on action buttons
- image uplaods / attachments with long filenames truncate with ellpsis
- attachments are not indented by an extra 15px in message bodies
Prevent progress bar overflow (was ending above 100%, 100.12315%...)
Update FileDownloadStore so it never creates Download objects when file is downloaded already
- Previously, the download itself decided if it would be a no-op, but this meant the download was around for a split second and you'd see progress indicators flash for a moment when opening/saving an attachment.
Upgrade FileDownloadStore use of promises
Restore Image attachment drag and drop - was broken because the name gradient thing was covering the entire drag region.
Allow file attachments to be drag and dropped to the finder and other applications 😍😍😍
Test Plan: Tests still pass
Reviewers: evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D1745
New In This Release:
- Reply/Reply-All keyboard shortcuts now update existing draft rather than creating another new one.
- Extensions can adjust new drafts before they're displayed
- Signatures extension! (signatures still need to be setup manually)
- Pristine drafts are discarded correctly when you leave a thread in two-pane mode
- Fix for "destroy draft" task failing
- Fix for "double sending" issue
- Upgrade to Electron 0.29.2
- New "Show / hide quoted text" control
- Bug fixes
Summary:
- Draft Store extensions can now implement `prepareNewDraft` to have an opportunity to change a draft before it's displayed for the first time.
- When composers are torn down, they delete their draft if it is still pristine. This makes the behavior of closing unedited popout drafts the same as leaving unedited inline drafts.
- The DraftStoreProxy keeps the initial body of the draft *if* it started in a pristine state. This means "is the body empty" is just a simple == check, and it takes into account anything added to the body by extensions.
- Calling Actions.destroyDraft doesn't blow up anymore if the draft session can't be found. This was a bug and meant that you couldn't destroy drafts which hadn't been previously edited, and also meant that bad things(tm) happened when you called destroyDraft twice, which seemed like overkill.
- DestroyDraft task now exits gracefully when the draft cannot be found.
You can test this feature by adding the following to your config.cson:
```
signatures:
NAMESPACEID: "<br/><br/><div id=\"Signature\"><div id=\"divtagdefaultwrapper\" style=\"font-size:12pt; color:#000000; background-color:#FFFFFF; font-family:Calibri,Arial,Helvetica,sans-serif\"><p></p><table border=\"0\" cellpadding=\"0\" cellspacing=\"0\" width=\"450\" style=\"font-family:'Times New Roman'; table-layout:fixed\"><tbody><tr><td class=\"logo-td\" align=\"left\" valign=\"top\" width=\"76\"><p style=\"margin-bottom:10px; margin-right:10px; font-family:Helvetica,Arial,sans-serif; font-size:14px; line-height:16px\"><a href=\"http://www.nylas.com/\" class=\"clink logo-container\" style=\"text-decoration:none\"><img alt=\"Nylas\" border=\"0\" class=\"sig-logo\" height=\"80\" width=\"66\" style=\"-webkit-user-select: none;\" src=\"https://s3-us-west-2.amazonaws.com/nylas-static-assets/nylas-email-signature.png\"></a></p><p class=\"social-list\" style=\"font-size:0px; line-height:0; font-family:Helvetica,Arial,sans-serif\"></p></td><td align=\"left\" valign=\"top\" nowrap=\"nowrap\" class=\"spacer-td\" width=\"16\" style=\"border-left-width:2px; border-left-style:solid; border-left-color:rgb(30,162,162)\"><img width=\"10\" style=\"-webkit-user-select: none;\" src=\"https://s3.amazonaws.com/htmlsig-assets/spacer.gif\"></td><td align=\"left\" valign=\"top\" nowrap=\"nowrap\" class=\"content-td\" width=\"368\"><div class=\"content-pad\"><p style=\"font-family:Helvetica,Arial,sans-serif; font-size:14px; line-height:16px; color:rgb(33,33,33); margin-bottom:10px\"><span class=\"txt signature_name-target sig-hide\" style=\"font-weight:bold; display:inline\">Gleb Polyakov</span> <span class=\"email-sep break\" style=\"display:inline\"><br></span><a class=\"link email signature_email-target sig-hide\" href=\"mailto:gleb@nylas.com\" style=\"color:rgb(30,162,162); text-decoration:none; display:inline\">gleb@nylas.com</a><span class=\"signature_email-sep sep\" style=\"display:inline\"> / </span><span class=\"txt signature_mobilephone-target sig-hide\" style=\"display:inline\">404-786-4100</span></p><p style=\"font-family:Helvetica,Arial,sans-serif; font-size:14px; line-height:16px; margin-bottom:10px\"><span class=\"txt signature_companyname-target sig-hide\" style=\"font-weight:bold; color:rgb(33,33,33); display:inline\">Nylas</span> <span class=\"company-sep break\" style=\"display:inline\"><br></span><span class=\"address-sep break\"></span><span class=\"address2-sep break\"></span><span class=\"website-sep break\"></span><a class=\"link signature_website-target sig-hide\" href=\"http://www.nylas.com/\" style=\"color:rgb(30,162,162); text-decoration:none; display:inline\">http://www.nylas.com</a></p></div></td></tr><tr><td colspan=\"3\"></td></tr><tr><td colspan=\"3\"></td></tr><tr><td colspan=\"3\"><p class=\"txt signature_disclaimer-target\" style=\"font-family:Helvetica,Arial,sans-serif; color:rgb(33,33,33); font-size:9px; line-height:12px; margin-top:10px\"></p></td></tr></tbody></table><p></p></div></div>"
```
specs for draft store extension hooks, some draft store refactoring
Test Plan: Run a few new specs that make sure extensions are run
Reviewers: evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D1741
Summary:
When bcc or cc is added, the ComposerView should show the fields. We should never hide a populated field!
Move the logic for determining message reply / replyTo participant sets into the message object. This seems OK because the functions don't modify the message and deal entirely with message attributes
Fix miscelaneous scrollbar issue
The MessageList does not always fire a composeReply action in response to keyboard shortcuts and can update the existing draft instead
There are a couple goals here:
- If you have an existing draft, command-R and command-shift-R should never create a new draft, they should just update and focus the existing draft.
- If you actually press the Reply button on the message item it should still create a new draft. It's still possible to have two.
- If you press Command-R and add a participant, and then press Command-Shift-R, it should add participants but not blow away the one you've added manually.
- If you press Command-Shift-R and then Command-R, it should remove the participants that were part of the "reply all", but leave other participants you've added untouched.
Test Plan: Run new tests!
Reviewers: evan
Reviewed By: evan
Maniphest Tasks: T2140
Differential Revision: https://phab.nylas.com/D1734
Summary:
Saving only 50 models at a time instead of 100 fixes a reproducible issue with database errors during initial sync on my machine. With 100, it reports "unexpected ; on line 2123..." during the query.
Only the main window should update the unread badge count.
Test Plan: Tests still pass and it seems like everything is still functional.
Reviewers: evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D1736
Summary:
Fixes T2275
Messages were still double sending sometimes because it took a while for
the Task to end up on the TaskQueue, which is what we were using to
determine re-sendability
Test Plan: edgehill --test
Reviewers: bengotow
Reviewed By: bengotow
Maniphest Tasks: T2275
Differential Revision: https://phab.nylas.com/D1728
Summary:
Fixes T2272
If the label is too long, the unread icons will now ellipsis truncate the
text correctly instead of being pushed off of the side.
I made the max-width of the sidebar 17px wider to allow for Karim's french
"Inbox" translation.
Fixes T2266
Added some more protection to places where filenames could be blank when
downloading.
This was partially fixed by Rob's D1685
Fixes T2258
You can now finish selecting a participant by pressing `space`, but only
when the participant looks like an email address.
When you do this we directly add the email address to the chip instead of
looking up a contact. This allows you to send just to the email instead of
adding a potentially erroneous name into the input field
Test Plan: edgehill --test
Reviewers: bengotow
Reviewed By: bengotow
Maniphest Tasks: T2272, T2266, T2258
Differential Revision: https://phab.nylas.com/D1729
Summary:
Fixes T2276
Since the new `QuotedHTMLParser` uses full HTML parsing, when you ask it
for the version of the document without the quoted nodes, it returns a
full html document. Unfortunately I forgot about the 2 <br> tags we
manually generate when creating a reply to a message with quoted text.
This is a pretty brittle solution, but considerably easier then checking
if the contents really are "blank". We could check innerText, but would
have to worry about img tags and other text-invisible-but-still-valid
elements. Since we control the generation of the reply in the
`DraftStore`, I think this is acceptable for now
Test Plan: edgehill --test
Reviewers: bengotow
Reviewed By: bengotow
Maniphest Tasks: T2276
Differential Revision: https://phab.nylas.com/D1730
Summary: Fixes T2179. This fixes the weird bug where first names like "The Nylas Team" was shortened to "The". The fix just chooses the first non-article word as the first name
Test Plan: tested manually
Reviewers: bengotow
Reviewed By: bengotow
Subscribers: jennie
Maniphest Tasks: T2179
Differential Revision: https://phab.nylas.com/D1715
Summary:
This diff does a couple things:
- Undo redo with a new undo/redo store that maintains it's own queue of undo/redo tasks. This queue is separate from the TaskQueue because not all tasks should be considered for undo history! Right now just the AddRemoveTagsTask is undoable.
- NylasAPI.makeRequest now returns a promise which resolves with the result or rejects with an error. For things that still need them, there's still `success` and `error` callbacks. I also added `started:(req) ->` which allows you to get the underlying request.
- Aborting a NylasAPI request now makes it call it's error callback / promise reject.
- You can now run code after perform local has completed using this syntax:
```
task = new AddRemoveTagsTask(focused, ['archive'], ['inbox'])
task.waitForPerformLocal().then ->
Actions.setFocus(collection: 'thread', item: nextFocus)
Actions.setCursorPosition(collection: 'thread', item: nextKeyboard)
Actions.queueTask(task)
```
- In specs, you can now use `advanceClock` to get through a Promise.then/catch/finally. Turns out it was using something low level and not using setTimeout(0).
- The TaskQueue uses promises better and defers a lot of the complexity around queueState for performLocal/performRemote to a task subclass called APITask. APITask implements "perform" and breaks it into "performLocal" and "performRemote".
- All tasks either resolve or reject. They're always removed from the queue, unless they resolve with Task.Status.Retry, which means they internally did a .catch (err) => Promise.resolve(Task.Status.Retry) and they want to be run again later.
- API tasks retry until they succeed or receive a NylasAPI.PermanentErrorCode (400,404,500), in which case they revert and finish.
- The AddRemoveTags Task can now take more than one thread! This is super cool because you can undo/redo a bulk action and also because we'll probably have a bulk tag modification API endpoint soon.
Getting undo / redo working revealed that the thread versioning system we built isn't working because the server was incrementing things by more than 1 at a time. Now we count the number of unresolved "optimistic" changes we've made to a given model, and only accept the server's version of it once the number of optimistic changes is back at zero.
Known Issues:
- AddRemoveTagsTasks aren't dependent on each other, so if you (undo/redo x lots) and then come back online, all the tasks try to add / remove all the tags at the same time. To fix this we can either allow the tasks to be merged together into a minimal set or make them block on each other.
- When Offline, you still get errors in the console for GET requests. Need to catch these and display an offline status bar.
- The metadata tasks haven't been updated yet to the new API. Wanted to get it reviewed first!
Test Plan: All the tests still pass!
Reviewers: evan
Reviewed By: evan
Differential Revision: https://phab.nylas.com/D1694
