performant-software / dm Goto Github PK

Color coding text, image targets, and text highlights; more standard text editor tools? (e.g. pt-driven font sizes)

Bugs, questions, requests that don't fit elsewhere.

This is from Martin Foys:

If you do a search on a term (say, "rome" in Virtual Mappa), and open a result, there's an issue. If you then click on the upper right corner page icon to see if anything is linked to it (which you would want to do in order to directly access the image of the map and target area linked to the annotation) - it will return "no annotations" IF the image it is linked to has not first been opened in DM. If the image has already been open in DM, then the upper right corner page icon will return the annotation's links to the image as it should. So it seems like if the image is not first opened, then the links to annotations from that image are not loading from the database if you open an annotation directly from search, as opposed to from the image it's linked to.

Martin:

Everything's been going swimmingly so far, but today (Saturday, about 12:00PM eastern standard time) it started taking a really long time for changes to the Bell-Tokens v.2 project to be saved. Creating a link between items is now taking several minutes to actually save (the status line display just reads "Saving . . . " during this time, and doing anything else during this just leads to sync errors and needing to restart the project. Sometimes, the change doesn't save at all, and just ends in "Not Saved - Sync Errors!" status display.

As backups are currently scheduled every 12 hours, this hints to a problem with the backup's transaction locking the triple store and hindering clients to commit changes.

Information Architecture Planning Space

• Planning one’s “Roadmap” through a project
• "Zoomed out" view
• Possibly pulling resources from categorized buckets onto a planning board and indicating coarse-grained relationships between them as a precursor to finer-grained annotation on contents of resources

On the “add a user” feature on the “Project Info & Sharing” window, would be helpful to simply have a way to display existing users that are not already on the project. Current window requires typing in the exact user ID, which means you have to know it first, and cannot find and select it.

On first opening a “public” display project, open the TOC window so guest users aren’t presented with a blank screen (so, symmetrical with what happens when a private user selects a project.

New issue: strings are not accurately being searched. In other words, a search string doesn't always return results which contain that string (Discussed with @gremid in May).

Search feature issue - search returns text items when some characters are searched for (like “” or “****”) that do not contain these characters in question. Conversely, searching for character strings like “[ ]” or “[“ does not return annotations in actuality do contain this string in their text.

Example: "????" as a query.

Also, partial strings are not matched -- might/might not be in scope for 1.0.0.

Options include:

• User Documentation, including how the syntax can be used.
• Change the parser, to make the search behavior more intuitive.

Having documents selected from the TOC load in the viewer to the right, not the left of frames currently open. The current scheme loads new docs to the left of current frames in the viewer, which means users have to think in reverse if they are opening up a set of files to work with, and have a specific order they wish them to display in.

Latest version of Firefox (48.01) does not display text files correctly. Text windows are empty. Clicked links from annotated images open a new window, but no content.

Perhaps this is intended, but if one follows a link from a resource to another resource that is already open, rather than going to the open resource, it opens a new read-only copy of that destination resource. That is especially odd when the read-only copy is not saved to the latest version of the open page, and the link may not even exist in the read-only copy, having not yet been saved.

Tags are not "atomic". If I drag one tag into the middle of another, they both break. If I shift a widget around in the text, the widget "identifier" may not be accurately moved with it, which means the tag no longer works on the page, but the resource to which it links will still bring you to this page.

Tags should be unbreakable via accidental insertion or deletion.

Logins are sticky. Even after I log out, when I open dm again, I am automatically logged back in, and cannot figure out a way to permanently keep me logged out. While convenient, not secure? Also, prevents me from testing public accounts, though this is more a personal issue.

• Should always log out on close
• Click logout, still shows you logged in
• Back button returns to page where user still appears to be logged in
• Blank page after logout redirect persists

The add user function allows one to add a blank user just by hitting the "+" button when there is no text in the "Add a user" field.

Remove the project-specific UI option of removing orphaned objects in the data store.

Then implement a server-side job, that periodically scans the store for orphaned objects

• logs them on the server
• optionally deletes them

Attempting to delete a project results in the following dialog:

How should the key data structures be represented? Possibly as a graph, since what's being built up over the course of the use of the application is a graph of documents interconnected by annotations.

Existing users cannot edit themselves, either their password or their email address. This is especially bad if their account was created using the Project Info and Sharing dialog (#7), so that now they can never edit their password or have an email address, and the username is forever taken.

Can we remove the “Clean Up Project” button from the TOC window - I believe it serves no purpose now?

Have a default option for bounding boxes and polygons that is clear on the inside with a colored frame.

Make it look nicer, and work better. This is not about creating per-user config functionality.

We've encountered an issue trying to rename an uploaded image file within DM:

In Virtual Mappa, there is a new map (recently uploaded) named "PsalterListMapReallyBig" - this is the name it had when it was uploaded.

We have several times tried to rename it "Psalter List Map (BL Additional MS 28681, f. 9v)" by editing the title bar above the image display.

When you edit it, the change appears to stick (and shows up changed in the Table of Contents as well) for the duration of the user being logged in, even if you change projects and then go back to Virtual Mappa.

But if you log out and log back in again, the title reverts back to its former, original state.

Martin:

Thumbnail loading for the Table of Contents display is extremely slow in some projects - can these images be rendered in lower file sizes to load more quickly?

I've had a text file disappear on me. It can no longer be seen in the Project window, but it's still in there somewhere because I can call it back up from a link, although in an earlier version.

If I click that link, it brings up the missing text file, although it is not saved to the latest version of that file.

Let me know if this is not clear enough. It's a weird situation.

Way to turn off the display of highlights on the text side, so it works the same way as turning off the targets on the image side

Autocomplete attempts to complete the search term with terms from other projects. If a term from another project is selected, the search provides 0 results.

Continuing bug: If you do a search on a term (say, "rome" in Virtual Mappa), and open a result, there's an issue. If you then click on the upper right corner page icon to see if anything is linked to it (which you would want to do in order to directly access the image of the map and target area linked to the annotation) - it will return "no annotations" IF the image it is linked to has not first been opened in DM. If the image has already been open in DM, then the upper right corner page icon will return the annotation's links to the image as it should. So it seems like if the image is not first opened, then the links to annotations from that image are not loading from the database if you open an annotation directly from search, as opposed to from the image it's linked to.

• User preference.
• Editors should be able to set rollover window size to a per-project default.

Description from Heather, via Martin. This supersedes my solution from December.

I worked on 18 inscriptions in the developer site (dm.performantsoftware.com) from about 4:50 pm to 5:10 pm (Central Standard Time) on December 21. For the first nine that I did, I entered my data so that the inscription would save. And it did. For the next seven, I entered the data so that the annotation would not save. And it did not. Then I switched back to the first method for the last two annotations and they indeed saved.

This is the difference between saving and not saving:

In both cases, the title changes usually save regardlesss. But...

1. if I start by making changes in the title, then make changes in the annotation text, the annotation does not save.

2. if I start by making changes in the annotation text, and then make changes in the title, the annotation will save.

[EDIT, 4/13/2017, from Martin]

Heather reports that she is occasionally still discovering annotations that have been made that are missing and have to be redone. Not a massive amount - only a handful, but that there still seem to be some annotation work that is failing to be saved.

From Martin Foys: A new issue on top of these issues has appeared - the last few highlights I have made and then linked from do not appear to be saving. I twice started up DM and made highlights on some text strings in the OE glossary, and then made links from them to highlights and targets on other texts and images. They took a loooooong time to save, but eventually reported as saved. But upon closing and reopening the DM project window, the last few highlights I had made were gone - THOUGH LINKS MADE FROM THEM ARE STILL VISIBLE FROM ROLLOVERS ON THE HIGHLIGHTS/TARGETS ON THE OTHER TEXTS/IMAGES.

In case it is helpful for zeroing in on the highlights that did not save, and the links to them that still exist, here's a composite graphic of them, so you can call them up yourselves - they are in the Bell-tokens v.2 project:

Ability to turn on an “editor mode” (or even just have it as a default if you have access to a project) where every item then open is already set to be editable, so you don't have to set editing for each doc you open (essentially having to open each one twice to work with it).

Have table of contents be ordered by an editor as opposed to being loaded alphabetically - click and drag structure?

Related to #31 : When editing, if you are in edit mode in a doc and you open a link to another file from it, could it open in edit mode as well? Reason: often when working on something, you follow a link to edit it, it’s in a long text file, opens up to the exact spot, and then you have to enable editing, and it resets the text display to the top of the file. Ideally, when you turn on editing for a file, it would open to the same place as you were in in the read only mode, and not reset to the top, but I’ve been told before that was not an option.

Clicking on an annotation that links to an image region, the linked region is not zoomed to anymore but the image is shown in total. Seems to be a regression bug.

We will allocate some time to creating some user interface mockups, which will express our recommendations for how DM v. 2.0 should look and work (based on the objectives, strengths and weaknesses of DM 1.0).

These mockups will be aspirational in nature. They will not make representations as to levels of effort required to implement the recommendations, nor will they be constrained by the implementation details inherent in DM 1.0. They will attempt to communicate one possible direction DM could take, in terms of user experience.

The intent of these mockups is to illustrate a compelling path forward for the project, which not only leverages the work done so far, but also takes advantage of new developments in web technology that have emerged in the time since work on 1.0 began, in terms both of user experience, and digital scholarship.

Search only operates on the content of text objects, not titles. It is not possible to locate via search an image or text by its title.

There’s no “saving . . .” alert bar at the bottom of frame for image files - only for texts. So this is asymmetrical.

Extend the export/ download of projects from a pure TTL export to one that packages the triples as well as referenced image data.

Allow also to restore projects by passing those export packages to a server instance on the CLI.

New users can be created simply by adding them to the "Shared" list on the Project Info and Sharing page. They can even be admins from this page. However, they will then have passwords the same as their username, and cannot ever change that or add an email. If it is desired that this be a place where users can be created, this function should call the Create user function if the name doesn't exist already.

When re-opening a newly created project, it appears to load, but a progress spinner runs endlessly in the center of the image. Even though prior annotations are visible and usable, the spinner never finishes.

Search returns on images that have the search term in the title - current search only returns text documents, not images with titles that contain the search term.

• Nice to have for 1.0.0

If one conducts a search and then switches projects, the search results are not cleared, allowing the user to summon objects from the previous project.

From Martin:

Text files will occasionally not load on initially opening a project - either sticking on a white frame with "loading" on it, or coming up as a gray frame (which is the first stage a text file opening appears as, I think). This is also happening with trying to open texts from an image link - where sometimes a link will get "stuck" and only open to a gray frame that can then not be closed. Sometimes this happens successively, but then will suddenly work again (see image - where five attempts to open the image link for to the Latin transcription for "scripturam" results in a gray screen, but then the final one (right next to the image) works.)

Switching to another the project leaves the gray windows on the screen (and, curiously, the version of the file that did open too - as in here (where I switched the project above to another, but these windows remained:

• Rollover windows for both text and image targets sometimes also get stuck on "finding annotations" - doing something else in the project (like going to other rollovers for other targets) usually fixes this stalling.

Have the ability to drop a uniform size icon on an image as a target (for instance, I'd love to be able to put before each line of text on a manuscript image the same exact size dot, but have no way to do that currently - it's impossible to create polygons, squares or dots of a uniform size).

When opening a text resource, a new window is opened but the text's content is not loaded, leaving the window empty:

Trying to close such empty windows triggers an exception:

Uncaught TypeError: Cannot read property 'uri' of null
    atb.viewer.ViewerGrid.removeViewerContainer @ all-js-code.js:767
    atb.viewer.ViewerContainer.close @ all-js-code.js:3210
    goog.events.fireListener @ all-js-code.js:445
    goog.events.handleBrowserEvent_ @ all-js-code.js:448
    (anonymous function) @ all-js-code.js:437

Empty windows also stay open when projects are switched.

From Martin FOys:

We discovered that if you try to upload an image that has a space in its file name, the upload will fail.

Do away with the distinction between "target" and "annotation" in rollover display of links for a given highlight/target. The original technical lead (pre-Performant) argued strongly for this, but from the user side of things, I just don't see the value, and have continuously fielded questions and confusion about it from other users.