SWB script frequently hangs browser

[Dave Stoft]

While precise evidence is difficult to gather remotely, one or more scripts frequently "hang" SWB -- not just when attempting to use the application but even just attempting to "look" at someone's plot data. The above screen-clip (Firefox debugger window) is just one example but anecdotal evidence over time (since the last SWB major "makeover") suggests the difficulty might be with "outside" scripts/widgets associated with "external" links with "unrelated tools and services"; i.e. with the likes of Facebook, Twitter, etc.

IF (bold intended) this were actually the "root cause" (or even A root cause) , then I'd suggest complete removal of such linkages as unnecessary interference with what, otherwise, was intended(??) to be a scientific application (right???). Links to "social media" probably do have benefits to other aspects of PLab, but I do question this specific linkage type to a measurement tool. [ On thinking about it, I'd still recommend removal of any and all "extraneous" linkages/tools/widgets/etc from the SWB interface page(s).]

Preface: I do not write this to complain but only to propose a new perspective; I do have other means of making spectral measurements and doing custom processing but the issues with SWB do make observing that data more difficult.

Yes, I will also admit that my experience with poor to zero performance with SWB may be worse than average, performance is likely related to distance to server and network traffic, and it is generally true that these types of issues will fall into some "bell curve" of user experiences. However, it is also a common characteristic that when a tool fails, a very small fraction of users will report it. The corollary is, therefore, that such an issue is generally much more significant than indicated by the # of times it is reported.

Because SWB is, in theory, a scientific, "cloud-based" measurement application, a primary criteria should be clutter-free intuitive efficiency and, therefore, the user interface should be optimized with that perspective in mind. IMHO, the most recent incarnation of SWB diverged completely from such a goal; despite/including the addition of new features. [ My guess is that much of this may be related only to the UI layer, not necessarily the underlying engines, which might make the solutions more tractable.]

I hope these observations are helpful.

Cheers,
Dave

[Jeffrey Warren]

Hi, Dave - thanks for the feedback. Can you link to pages in particular that show slowness? There is an open issue for optimizing processing time in SWB, which may help, but I'm interested to hear that the outside widgets may be causing trouble. 

Indeed there's no reason they should, and I believe we could replace them with passive links to share data, rather than their relying on an additional request, which seems silly (sadly that's the standard way to add such links, according to Facebook and Twitter, but we don't have to follow their examples)

I've also noticed that some people may add many redundant and overlapping operations to a spectrum, adding a great deal of overhead which has no analytical function, and which may cause extreme slowness -- for example multiple linearCalibration operations. I've tried to walk a fine line with /forbidding/ operations which don't make any sense (it's frustrating to have a program tell you not to do something!) but in this case I think we should throw an error rather than let the program run more than one stacked calibration -- it should say "remove the existing calibration to add a new one" or something. I'm making an issue for that too now. 

Thanks for your detailed input -- bugs never get fixed unless someone reports them!

Jeff

--
Post to this group at plots-sp...@googlegroups.com
 
Public Lab mailing lists (http://publiclab.org/lists) are great for discussion, but to get attribution, open source your work, and make it easy for others to find and cite your contributions, please publish your work at http://publiclab.org
---
You received this message because you are subscribed to the Google Groups "plots-spectrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plots-spectrome...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jeffrey Warren]

Issue on spectra with redundant tags: https://github.com/publiclab/spectral-workbench.js/issues/48