Feedback

The Feedback row of the Account page and the report button of the title bar, reporting a bug, suggesting an improvement or requesting a driver with a capture of the screen, under the station's Ganter account or without one, what is sent, and when it cannot be sent.

The Feedback row, under the SUPPORT group of the Account page's index, is a short form that sends a report to Ganter. The same form opens over any page from the report button in the title bar, beside Help, whose tooltip names it Send feedback, and there it starts with a capture of the screen you were looking at (see The report button). It is one form in both places: a report begun in the dialog and closed unsent is the report this row shows, capture included. It has three kinds, one required line, one optional text and the screen capture. A station signed in to its Ganter account sends under that account, so Ganter can reply to its e-mail; a station that is not signed in sends too, without an account, and Ganter then has no address to reply to (see Without an account). Who may send is a permission of the role, Send reports to Ganter, on in every role until an administrator turns it off (see Permissions).

When it cannot be sent

A role that may never send is not shown the button at all, because no press would earn it the permission; a situation that passes keeps the button where it is, greyed, with the reason as its tooltip, under the button's name. The published demonstration draws no report button, as it draws no Account page. A station that is not signed in to a Ganter account is none of these situations: the form shows and sends (see Without an account). A station signed in sends under its account or not at all, online or offline: sending while offline tries to renew the account's token first, and when that fails the send is refused with "The station's Ganter account could not be confirmed for this report. Sign in again on the Account page and send it again." rather than going without the account.

Without an account

A station that is not signed in to a Ganter account sends its reports all the same, with the same fields and the same capture. Above the fields the form says "This station is not signed in to a Ganter account, so the report goes without an e-mail address and Ganter cannot reply to it. To get a reply, sign in to the station's Ganter account on the Account page." The form asks for no address of its own. Signing in while a report is being written takes the line away, and the report then goes under the account.

The report button

The report button stands in the title bar beside Help, on every page: a glyph alone, the size of the Help key. Its tooltip names it, Send feedback, and says on the line under the name what it does, or why it is greyed; a screen reader announces it as Send feedback. A press does two things, in this order:

1. It captures the window as it is at that moment, through the same capture the agent's ui_screenshot tool takes. While it is being taken the button is greyed with "Taking the screen capture."
2. Only then does it open the form in a dialog over the page, Report to Ganter, with the capture shown under Send the screen capture, ticked.

The form never appears in its own capture, because the picture is taken before the dialog opens. The dialog holds this row's form, with Close beside Send; Escape and a click outside it close it too. Closing it sends nothing and loses nothing: the next press, or this row, shows the report as it was left, and a new press replaces only the capture. A report the server takes closes the dialog.

A report drafted by an agent

An AI agent connected to the station through agent access can write a report for you with its report_draft tool (see Reports: draft): a driver Ganter Lab does not have, with the make, model and protocol it tried, a defect of the program, or an improvement you asked for. It opens this form the way the report button does: the window is captured first, then the dialog opens over the page with Type, Summary and Details filled in and the capture ticked. Above the fields the form says "An agent connected to this station wrote this report. Read it and correct anything before you press Send: nothing goes to Ganter until you do." Under the capture it says "The screen as it was when the agent opened this report. It shows everything that was on it; untick the box to send the text alone."

The agent never presses Send. You read what it wrote, which can name devices, addresses and values of the plant, then correct it, untick the capture, close it or send it; the outcome reaches the action feed as every report's does. The Events console keeps the draft under the Agent area and, once you send it, a line saying that the report the agent drafted was sent.

The form

The hint under Summary follows the type:

Under the box the form says "The screen as it was when you pressed Send feedback. It shows everything that was on it; untick the box to send the text alone." and shows the picture itself, which is exactly the file that leaves. What else the form may say about the capture:

How a large capture is made to fit

The server keeps a capture of at most 4 MiB (4,194,304 bytes) of PNG whose image is at most 40 million pixels. The station checks both before the form shows the capture. A capture past either is reduced: its width and its height are halved, each block of two by two pixels averaged into one, and it is encoded again, until it fits. A full HD window fits as it is; a 4K screen showing a busy page may be halved once. A capture that would have to be narrower than 320 pixels to fit is left out instead, because nobody could read it.

On success the feed reads "Thanks! Your "" was sent." and both text fields clear, unless their text changed while the report was on its way: then it stays, and the thanks still names the type that was sent. The capture clears with the report it belonged to, ticked or not. The server keeps at most a few captures a day; past that it keeps the report without its picture, and the feed says so calmly: "Thanks! Your "" was sent without its screen capture: the server already holds as many captures as it takes in a day." On failure the feed carries the reason and the text and the capture stay for another try:

What is sent

The capture shows everything that was on the window, the plant's names and values included, which is why you see it before it goes and can untick it, and why the role permission decides who may send one at all. Ganter keeps it with the report and mails it with the report's notification; it is deleted with the report, and, for a report sent under an account, when that account is closed. Nothing else is attached: no log excerpt, no configuration, no device list, no local user name and no support contact. What you want Ganter to know goes in the text. A crash is a different kind of report, captured automatically and sent only with your per-crash consent from the dialog described under Crash reports; that one may carry the crash details, may be sent anonymously, and never carries a screen capture. The Events console can produce a support package, on the Console tab, if you need to hand over diagnostics alongside a report.

What the row does not do

• It does not queue a report for later. A send that fails is refused at once; the text and the capture stay in the form until you send again.
• It does not answer back. Replies, if any, reach the account's e-mail, not the app; a report sent without an account gets none.
• It does not ask for an e-mail address. Without an account the report goes with none.
• It does not send from a browser or on the demo profile.
• It does not capture the screen by itself. The capture comes from the report button, of the page you were on when you pressed it, and the station keeps no file of it.
• It does not attach diagnostics of any kind, and it does not send the runtime state.
• It does not let an agent send. An agent can open it filled in (A report drafted by an agent), and only your Send sends it.