Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Could qplot be on SSC? On SSC and SJ program updates

    Dear all,

    I am facing a particular issue, but I'm sure the general explanation of it could be useful, to me but to many other also.

    By chance, I read a post (p.10/20) on qplot program developed by Nick Cox (SJ 12(1):167 for last update).

    This program published in the Stata Journal found not be hosted in SSC (as I thought it was the case for program published in the Stata Journal, and as Nick uses to do).
    Hence to install it I had to type search qplot, open the package gr42_4 window and click on "click here to install". (the quicker ssc install qplot returned ssc install: "qplot" not found at SSC, give a try if you want to.)

    Ok, not so hard you'll tell me. However, I have to work most of the time in a "secured" (offline) distant server, to prevent from confidential data leaks (the CASD for those who know it).
    Our administrators kindly understand this situation was very tedious, notably to update our favorite statistical software, so they automatically download the SSC repertory, which allow us to install them finally.

    Long story short : qplot not being on SSC, I can't download it (try to click on "click here to install", when being offline you'll see the error message I got).

    When it becomes interesting is that I saw that qplot is the new name of an old program (quantil2), which turned to be available on SSC (and which I have downloaded on the "secured" server).

    This means that Stata Journal updates aren't stored in SSC? Or this only happens when the program name changes?
    Anyway, since I haven't written the program I won't upload it to SSC, but I find strange that SSC keeps online a (very) old program when (several) updates have been published in SJ.

    So I have two questions:
    1) Could anyone tell me how SSC precisely works, and with SJ published user-written software ? (only indicating a good link would be enough)
    2) What would be the best solution to draw something like a qplot without qplot (until now I draw CDF curves, but it also was tedious). Using quantil2 isn't a serious alternative (the output is very old-fashioned Stata)


    Nb : Of course I have sent an e-mail to our administrator to download this precise program, but I don't know how much time it will get to be treated. Moreover, I'd like to understand the whole process from SJ to SSC, for me (and others) to know what to do next time.

    Best,
    Charlie




  • #2
    Could anyone tell me how SSC precisely works, and with SJ published user-written software ? (only indicating a good link would be enough)
    I assume you do not have the technical details in mind? You can access the source code for Stata's ssc command.

    So, start with here and read more on how to submit programs to the SSC. There really is not much to it. Kit Baum invests his time and does all the hard work here.

    Visit the SJ Homepage to read more on how to submit programs to the SJ.

    In general, SSC and SJ work completely independent from each other. There seems not be a rule that programs submitted to either of the two cannot be (re-)published in the other, so this is entirely up to the program authors. I do not have any information whether an update to a program on the SSC requires an update to the same program if it was also published in the SJ.

    Best
    Daniel

    Comment


    • #3

      I think there are two main issues here.

      1. People living behind firewalls find that those firewalls complicate access to user-written programs. Nothing said below indicates lack of sympathy with that. Your firewall is yours to discuss locally. Serious Stata users may often want access to the Stata Journal files and to SSC (as well as to other sites).

      2. Lack of clarity on what is posted where. Here I have some responsibility, particularly in regard to my stuff and more generally as an Editor of the Stata Journal.

      Main facts: SSC has precisely no formal links to the Stata Journal. Whether authors post on SSC or not is at their discretion (not entirely at their discretion, as for example, Kit Baum won't post stuff without help files, or so I believe; nor would he (could he) post two or more packages or programs with the same name). That has always been true. The Stata Journal doesn't encourage, discourage, instruct, prohibit or inhibit authors posting or not posting their programs anywhere else they like, SSC, GitHub, their own sites, etc., either before or after publication in the Stata Journal.

      The Stata Journal takes seriously its archival function as providing a record of successive versions of programs. Empirically, most users of most programs just want the most up-to-date version, but that doesn't rule out or undermine the archive character. Conversely, SSC doesn't really have any idea of previous versions except and unless people rename their programs so that different versions have different names. (Fans of GitHub are free to underline that GitHub allows precisely this, and much more.)

      The SSC archive has a precisely limited status insofar as ssc is an official command easing access to it. No more, no less. After all, you can read Excel or SAS or some other kinds of files with Stata; that implies no more than it says. StataCorp doesn't control SSC in any sense whatsoever, even the Stata files on it.

      The rest of the post is personal opinion and/or specifically about my programs.

      qplot (mine) has never been published on SSC so far as I can recall. quantil2 is up there in a version from 2002.

      I am sorry that this is confusing. Equally, there is some caprice in where I post stuff and whether I submit a write-up of any programs to the Stata Journal, but others have precisely the same caprice, so I don't think I need to apologise much. A package of mine njc_stuff (which is on SSC) documents what is where. I never let its public version get seriously out of date.

      I do post extra programs from time to time within posts on Statalist but never promise that I will write all of them up elsewhere.

      The argument for keeping old stuff available anywhere is three-fold (or four-fold):

      1. It should do no harm. Stata provides several means to find what is available. People do have some obligation to check what they are using. If programmers are unclear (say by not including version numbers in their codes, or not maintaining clear help files), you should know what to think.

      2. It is impossible to be certain what is no longer useful to anyone anywhere. People are still using some very old versions of Stata and for good reasons whenever there just isn't the money to upgrade, or the inclination on the part of those in power over you.

      3. Sometimes people want to reproduce what was done with earlier versions of software. 10 years is a long time in software, but not in many disciplines.

      4. Personally, I have no interest in spending time worrying about any of the above (e.g. asking Kit Baum to remove very old stuff), once I've done my best to explain what of mine is what and where.

      Comment

      Working...
      X