T
Hi, I reduced this to a small example that may help explain it. Save this as example.js: define(function () { var nodeModule = {}; try { var types =nodeModule.require && nodeModule.require("util").types; } catch (error) {} return { loaded: true }; }); Then load it through Dojo: require(["./example"], function (example) {console.log(example.loaded);}); With factory scanning enabled, Dojo attempts to load util.js. The optional call would never execute because nodeModule.require is undefined, but…
T
Thanks for the update, @ajay. I got this one wrong for your case. ECO-1916 shows it intermittently on sites where the app was never installed, so neither install age nor the Connect migration path explains it. The manifest reference note that “existing permission configuration won’t be overwritten during add-on upgrades or re-installations” still holds for upgrades, but it isn’t the cause here. For anyone else who lands here, I’d watch and vote on ECO-1916. The workaround it gives is for a Jira…
T
Hi Atlassian developers! Cover Music and Copyright When a video on this channel features a MARK HAMILTON cover, I perform the music myself from sheet music and make my own recording. I also create all the drum and percussion parts myself, including the kick, snare, and hi-hats. Examples of my covers: John Williams – The Imperial March (MARK HAMILTON Cover) Pyotr Tchaikovsky – Dance of the Sugar Plum Fairy, from The Nutcracker (MARK HAMILTON Cover) Any applicable copyrights in the original…
T
Thanks! Glad it’s not just us. What we saw: the “Allow access” prompt only appears when the resolver call itself fails with NEEDS_AUTHENTICATION_ERR. As soon as our handle() wrapper caught every error and returned { ok: false, error }, the call “succeeded” from Forge’s point of view, so the UI just showed our error text and the consent prompt never appeared. Re-throwing only that error fixed it in all six affected apps: } catch (e) { if (e?.name === ‘NEEDS_AUTHENTICATION_ERR’) throw e; return {…
T
@SolomanWeng sorry, the problem is with the latest versions, too. I was experimenting with library.type: "umd" and kept this setting when trying the latest versions. This workaround works, too, for the old and the newest editor-json-transformer. So, there are two workarounds for my problem: either to use “umd” type or switch off concatenateModules optimization. Strangely, “umd” type seems to use concatenateModules optimization by default, too, and it’s working. Problem is, particularly, that we…
T
FilipIlic: 5 Forge traps the docs don't mention | Loopwise Labs awesome thanks Point 1 really interests me, since I ran into that as well. Can Atlassian confirm that pattern to rethrow NEEDS_AUTHENTICATION_ERR errors?
T
An update on my original report: Atlassian has acknowledged the issue through ECOHELP-178255 and created a separate public bug: ECO-1916 — Forge globalPermission defaultGrants intermittently fails to apply on installation. Thanks to everyone who shared suggestions here. The confirmed bug affects fresh installations, even with a single jira:globalPermission module configured with defaultGrants: all. The permission is created, but its default group grants are intermittently missing. This goes…
T
Hey @BenRomberg, sorry it’s been a while! In case you didn’t see this elsewhere, we have an Early Access Program (EAP) you can use to try out multiple contexts per space today. Just raise a request here and we can add your test tenant to the EAP. However, this feature is now rolling out to all customers over the next few weeks, so you should see it in most production tenants soon.
T
For our Confluence app conversion we moved from AP.cookie to directly using local/sessionStore in Forge (customUI), with a key that included the cloudId and accountId. The props were low risk (e.g. have I read the latest news bulletin), so we didn’t have to migrate data.
T
Hi @RiturajSeal Screenshots below. It’s better than before, but not pixel-perfect. The yellow padding from a diff outside the iframe is too big. As shown in the second screenshot, the content in the iframe doesn’t align with the app logo, which I think was the intention.
T
Thank you for your patience on this @goyoregalado, and apologies we have not post follow up sooner. We do plan to go ahead with this change, but we decided to defer it from Q1 (Sept-Dec) to Q2 or later. We haven’t yet committed to a specific timeframe. We have a few other features we are considering that tie in loosely to this, so we want to make sure they all fit together: We’re investigating a Requires Review status (similar to Incomplete on Connect migrations) that is somewhere between a…
T
Thanks David. We checked this on a test site. With the dashboard filtered by space, status and priority, the built-in “Work item list” widget returned exactly the work items the equivalent JQL returns, every time. The metrics model the dashboard loads also maps 23 Jira filter dimensions to JQL fields (project_name to project, issue_status to status, and so on). So the platform already has what apps need. Either the filter as JQL or the selected work items in the widget context would solve it;…
T
@AllanahMiller two questions from the DashboardMigrationPayload type published in @forge/dashboards-bridge 2.0.3, both about a plain copy-through transformer: source.dashboardItem.properties is typed Record<string, string | boolean | number>, but our gadgets save their configuration as a nested JSON object under one property key. Will object-valued item properties arrive parsed, or as JSON strings? A copy-through transformer depends on knowing which. In #16 you mentioned the transformer input…
T
Hi @AlexanderSavovski The reason why Atlassian is not exposing the filters as JQL is because these filters are running against the Atlassian data lake, not the Jira database. While you might be able to convert some filters to a JQL equivalent, the results are never going to be the same as what built-in widgets will show, if nothing else because of data lake consolidation latency. Also, the data lake is not built to return individual Jira work items but aggregated results.
T
While working with the Get item REST API call using curl, getting the following response for items I can work with via the standard Jira Cloud web interface. It feels like I’m missing something basic, and any assistance would be greatly appreciated. {"errorMessages":["Issue does not exist or you do not have permission to see it."],"errors":{}} I believe to have successfully created a “classic” API token following the provided documentation. Using the same curl call to the “Server info” endpoint…
T
Hi dashboards team, Adding our case to Raimundas’s question above (post 28), also raised with developer support as ECOHELP-183858. Our reporting widgets count work items with JQL. When a dashboard is filtered, Atlassian’s own widgets follow the filter bar, but an app widget receives the filter only as extension.filters, an undocumented structure with no JQL. Apps can apply only the filters they manage to reverse-engineer, so the same dashboard shows conflicting numbers, and customers read that…
T
Hi @BayuTrisnapati, I reproduced it on our test site yesterday morning. An app account was behind it each time. The unions only allow AtlassianUser: union ConfluencePageOwnersResponse = AtlassianUser A page owned by a Forge app’s account gave edges[0] null with your exact error, for owners, contributors and creator. Contributors on a page with six authors returned five. By elimination, the null was its app account. The other edges still came back, so skip null edges and fall back to the REST…
T
Hi @NeelDedhia thanks for the update. For us, the spacing issue is not fixed. There’s still 8px of left padding (if disabled, other CSS kicks in and adds 19px of left padding). The border radius issue in the header is still present when viewing the work item in a pop-over/center view. Appears to be fixed in the full-page view and the sidebar. Please install our app or create a sample app to verify your own changes.
T
AllanahMiller: Alexey (if you read this), could you expand on your need for advance notice of site conversion, including what your app needs to do beforehand and how much notice you would need? Hi Alannah, thanks for the clarification. To answer your question: if we create the transformers ourselves, then we likely won’t need advance notice regarding the migration.
T
Hi @AndrewMiller1 , According to the RFC timeline, the resolution was expected by October 15. Is there any update on the outcome? Has Atlassian decided whether to move forward with the proposed custom migration progress APIs? If the proposal has been accepted, is there a public roadmap or an estimated timeline for implementation? Thank you in advance!
T
nathanwaters: So in addition to prefetch they probably also need to be pre-warming those lambdas. @HeyJoe just checking in with you, as I would consider this related to RFC-152: Increasing maximum memory for Forge functions In GCP Cloud Functions, we have the ability to configure a minimum number of “ready” instances: Functions best practices | Cloud Run | Google Cloud Documentation These instances are always kept warm, allowing applications to minimise the impact of cold starts. The platform…
T
Hey, We are waiting for a long time to be able to set visiblity of forge defined customfields types (however our fields are of this types: group, user, number, string). We need in first order this easy functionality for hiding them (displayConditions are not enough for our case). Second thing is Global Issue Create screen, We would rly like some way of reacting for “create button”, i mean the ability to send with it some properties based on forge customfields. We are using it on ServiceDesk…
T
@VickyHu thanks for the proposal. Still processing it - meanwhile, it looks like the link to https://forge-a2ui-catalogue.kaizen.shared.atlassian-3p.com/ is only returning Site not available.
T
Hi dashboards team, Exporting a new dashboard to PDF (⋯ → Export → PDF) shows every Forge app widget as a broken link card (“We couldn’t load this link for an unknown reason”) instead of the widget. Atlassian’s own charts export fine. We reported it with steps and screenshots in ECOHELP-183631. Is exporting Forge widgets planned, or an API for a widget to supply content for the export? Until then, a neutral “this widget can’t be exported” placeholder would read better than the raw iframe…
T
Correct me if I’m wrong but I’m pretty sure the Forge platform is built on AWS Lambda. That means invocations have the cold start problem, which is why everything on Forge is noticeably MUCH slower than Connect. They should have gone for something that doesn’t have that problem like Cloudflare Workers. And then they simply slap a comically-oversized loading spinner onto every module surface without bothering to vertically center it So in addition to prefetch they probably also need to be…
T
hi, @MattGauntSeo thanks so much for reporting it, we have confirmed it is an issue and will create a patch for it. thanks
T
Following the release of Forge Rovo agents, Rovo MCP, and Rovo actions, we’re exploring how your Forge agents and tools can present interactive UI on Atlassian Rovo surfaces (e.g. Rovo Chat) and in external AI experiences. This RFC seeks early feedback on a proposed Early Access Program (EAP) for A2UI components in Rovo actions via Rovo Agents rendered in Rovo Chat. It also outlines our longer-term direction: making those components available through Rovo MCP (Preview) in external AI chats,…
T
In @atlaskit/editor-markdown-transformer > media.js, the media rule looks back through the input token list to find the opening tokens, then uses that index to cut the output list. After the first image, the output list is longer than the input list, so each later image cuts the output in the wrong place: the parser crashes (TypeError: Cannot read properties of undefined (reading ‘type’)) or text is silently lost. The rule also runs before inline parsing, so image syntax inside a code span is…
T
super valuable feedback thank you! I will take this to the team. AGC - not available yet, some architectural blockers are currently being worked on to allow this to happen soon. I don’t have a timeline yet unfortunately but I’ll update you here if I hear anything. Isolated cloud - yes, dynamic modules are available
T
@NeelDedhia Any updates on this?