How would you improve the File Exchange site?

In sharing my codes on the Mathworks File Exchange (FEX) site, I've run into a few usability issues. My activity on the FEX site tends to ebb and flow, and every time I return from a lengthy hiatus I think, "Wow, all this time and they still haven't fixed the problem with _______?"

How would you like to see the FEX site improved? I'll get the ball rolling with a few of my own ideas.

Answer 1

[Jan]

While I think that the submission process is fast enough and problems, which does not match in the dull input mask can be solved by a personal contact to the editors, I actually agree, that the FEX can be improved.

I still struggle with the license files: Because all license files have the same name, I cannot store several FEX downloads in the same folder without collisions. Adding a folder to the path for each submissions is not efficient also. Overwriting the license files, because they are almost equal at all, is not valid, because this is almost the only thing, which is forbidden by the license. I'd prefer a link inside the M-files to the BSD-template file in the net.

I agree that minor changes in the documentation should be easier than the full re-submission process.

Editing the comments as in the forum would be helpful also.

There are much to few ratings and comments. This feature would be such powerful if if is used frequently.

But in total I really like the improvements I've seen in the last years: The new list style, the high-lights block on the right, the BSD-license, animated GIFs.

Answer 2

[Chad Greene]

I'd like to see trusted contributors given the ability to publish files more rapidly. Say we have another big event like a major oil spill or a volcanic eruption and a clever File Exchange contributor comes up with a slick way to import, display, or analyze new, relevant data using Matlab. If the file approval process takes a week, it may be a bit late for other scienists to build on this work.

[Walter Roberson]

In a case such as that, post your code on some filesharing site and post links to it in Answers and the newsgroup. FEX is intended for polished tested code anyhow, not for rushed code.

[Chad Greene]

Walter,

I wholeheartedly disagree with the limitations you've placed on the purpose of File Exchange. And rushed code can absolutely be accurate and relevant code that helps push science forward.

Chad

[Walter Roberson]

Are you suggesting that FEX is the place for any MATLAB code, no matter how unpolished and untested?

[Chad Greene]

Hi Walter,

Yes, FEX must welcome all contributions without discrimination based on the skill level of the author. It's a User Community, not an exclusive club. I have personally benefited from unpolished code that others have shared on this site. The Pick-of-the-Week badge may indicate the level of quality you're looking for; but even more valuable, user reviews and comments do so with more nuance than an administrator's simple approval or disapproval of a script. Good work gets recognized and widely implemented, while unpolished work is identified, giving the author a chance to improve it. This is the basis of peer review--and I would like to see FEX allow for this peer review to occur on 21st century timescales.

Chad

[Jan]

I've posted about 40 files in the FEX and updated each of them about two times. I never waited for more than 2 business days until they appeared in the FEX, and the average time was about 24 hours.

So I agree: "If the file approval process takes a week, ...". But I'm convinced that it does not take a week.

I do not think that important, not to say urgent, code concerning the analysis of disasters should be shared using the FEX. This is simply not its purpose. Imagine the nuclear power plant in your neighborhood crashes. Which serious scientist would go to the FEX and checks, if there is a corresponding submission already?

[Chad Greene]

Jan,

In the days following the Deepwater Horizon oil spill, I remember checking FEX to see if anyone had posted any related data-grab functions or analysis scripts. A simple data import function could be a launching pad for the scientist who wants to analyze data, but does not have the time or expertise to figure out how import the data into Matlab.

In the first few weeks of the spill, every major funding agency received a deluge of research proposals that had been written in record time. These proposals built their cases on preliminary data because funding needed to be delivered quickly.

If I had written a data-grabbing function the day after the explosion, then noticed a typo in my script immediately after uploading, we could have to wait a day or two for the file containing the error to be approved by FEX. Then I'd click "Update File", upload the correct version, and wait all weekend, when scientists were frantically working on this problem, for the correct file to be approved. This is a case where an error in an uploaded file could be noticed on a Wednesday, but the corrected version of the file would not be uploaded until nearly a week later.

This case is surely rare, but not entirely hypothetical. In fact, just this past weekend I uploaded a file, immediately noticed an improvement that I could make, and I am still waiting for the new file to be approved.

Chad

Answer 3

[Chad Greene]

The site tends to be rather slow. It's about as responsive as a 1990s gif-laden forum for antique tractor enthusiasts. This should be somewhat embarrassing for a modern software company that touts its products' computational speed and utility in cutting-edge research.

[Jan]

I agree that the FEX is not a high-speed page. After I had to buy a new computer to get more than 1 key-stroke per second in this forum, I think the speed of the FEX is not the main problem of the bandwidth of the TMW servers.

Answer 4

[Chad Greene]

There is no way to confirm that a submission has been uploaded. When I upload a new file, I receive no confirmation email, and I don't see the new file on my File Exchange page. In my list of posted files, a simple grayed-out file name would suffice. But ideally, I'd like to be able to visit the preview page until the submission is approved and goes public.

[Jan]

When I hit the "submit" button, a page appears which confirms the submission and explains the process of approvement. Because a submission never failed for me, I think this level of confirmation is sufficient.

Answer 5

[Chad Greene]

There's no way to edit files after submission. Recently, I uploaded a new file on a Thursday. A few minutes after I clicked "Submit", I realized that my script contained a typo in which two variables were multiplied where they should have been subtracted.

"Yikes!", I thought. I quickly zipped up a corrected version, but I could not update the file until after the submission had been approved by the FEX administrator. My file went public on Friday, and within minutes of its approval I clicked "Update File" and uploaded a new version. Unfortunately, the updated version remained in limbo all weekend and I had to watch the download count steadily increase as users downloaded my incorrect file.

[Jan]

You can simply contact the FEX admins by email. I've done this repeatedly and got a fast response. Just look at the "Contact us at ..." line on the file submission page. This is not an empty phrase, but TMW members really reply to emails. For your case an email with the improved version as attachment and a clear explanation should be an efficient and working strategy.

[Chad Greene]

Jan, I'm glad to hear of your successes contacting the good folks at File Exchange, but I have not had such luck. All of my messages to them have gone into an abyss--I have never received any response. I suspect you may benefit from name recognition, but all the nobodies get ignored.

[Jan]

I admit, the fact that you did not get any response to all of your mails, sounds clearly like my very positive impressions about the quality of communication are strongly biased. On the other hand, I had the first positive experiences with TMW members before I participated in the forums, e.g. in 2004. Therefore the name recognition is not plausible. But then I cannot explain, that out experiences seems to be such different.