Catalogue not displayed on nightly

[jykae]

Closes #1214

Proposed changes

• check bookmarks result, user might not have bookmarks, which results undefined.
• add reactivevar import
• paranoid check user query, in case of undefined

[jykae]

Tried deploying this branch for temporary server to test in production one exception occurring on nightly but not in development.
However I could not catch this error on development and not in test deployment,

TypeError: Cannot read property 'splice' of null
    at https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:107:29128
    at Object.o.nonreactive (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:52:4311)
    at Object.removedAt (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:107:28856)
    at Object.removed (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:103:1831)
    at https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:70:1323
    at Array.forEach (native)
    at Function.A.each.A.forEach (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:1:865)
    at r.diffQueryOrderedChanges (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:70:1281)
    at l (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:103:1417)
    at https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:103:802

Catalogue and each tab opens ok for me in development and test production deployment with these changes. Could someone of @apinf/developers do review?

[55]

Getting on nightly:

2a2fb88….js?meteor_js_resource=true:3 Exception from Tracker recompute function:
2a2fb88….js?meteor_js_resource=true:3 TypeError: Cannot read property 'username' of undefined
    at https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:428:4436
    at Array.map (native)
    at Function.A.map.A.collect (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:1:1117)
    at t.e [as getApiManagersByName] (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:428:4386)
    at https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:126:2518
    at s.call (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:126:1929)
    at s.mustacheImpl (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:126:1184)
    at Object.s.mustache (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:126:1243)
    at ._render (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:440:29692)
    at l (https://nightly.apinf.io/2a2fb884adc53dffc6e76ae838b2444f03baafbd.js?meteor_js_resource=true:107:20600)

[55]

@jykae, if it's working on test deployment, let's merge it and do a manual deployment to nightly, since the catalogue is a big thing.

[jykae]

@NNN Ok, I think "username" is now checked as much as one can. I could manual deploy on nightly if you merge this.

[jykae]

@NNN Deployed this branch on nightly, https://travis-ci.org/apinf/api-umbrella-dashboard/builds/140729834

Check https://nightly.apinf.io/catalogue

[jykae]

As you notice there's one weird edge case, where this "if-statement" does not happen, https://github.com/apinf/api-umbrella-dashboard/blob/bfa7d9475fc0ea0f7bb02cb97401bff5db91059a/apis/collection/helpers.js#L129

@NNN Is it ok to return nothing like it is now, or should we return there also "admin"? What you think?

[55]

@jykae, sounds right for me. @brylie, what do you think?

[brylie]

This should also be moved inside of the if (userBookmarks) { .. } block.

Also, wait until @NNN merges changes in pr #1221, as some of these changes overlap. Specifically, changes to the publications.js are duplicating changes in PR #1221.

[brylie]

This PR looks good. Lets hold off until we get PR #1221 merged, and see how that relates to this PR. I.e. the changes to publication.js may not be necessary after the PR #1221 merge.