[ESS] tracebug ess-inject-source race condition

[Martin Maechler]

>>>>> Vitalie Spinu <spinuvit at gmail.com>
>>>>>     on Mon, 24 Jun 2013 03:25:47 +0200 writes:

    > Thanks for reporting this. 

    > I guess the only reliable solution is to not delete temp files at
    > all. Or at least do the cleaning only on the process termination. 

    > Vitalie

We definitely should keep a clean up procedure.
"on process termination" would be nice, but could that really
happen reliably?

A well working alternative might be to clean up files which must
be named in our "file pattern" if they are older than one week
or so.
{may harm people who run an ESS for longer than one week, but
 should they really be doing that ??}

Martin


    >>> Peter Meilstrup <peter.meilstrup at gmail.com>
    >>> on Sun, 23 Jun 2013 14:41:20 -0700 wrote:

    >> Here's a problem with how tracebug evaluates code chunks.
    >> (1) set ess-inject-source to t

    >> (2) Create an .R file with some code chunks that take an appreciable time
    >> to evaluate. Then send a series of chunks using C-c -C-c (or faster: C-c
    >> C-c, C-x C-z, z, z, z, z, z). The point is that ESS is sending chunks
    >> faster than the R process runs them.

    >> This results in ESS deleting temporary files before R can see them,
    >> resulting in errors like:

    >>> Error in file(filename, "r", encoding = encoding) :
    >> cannot open the connection
    >> In addition: Warning message:
    >> In file(filename, "r", encoding = encoding) :
    >> cannot open file
    >> '/var/folders/Zz/ZzJ1nC8pHNGtZb-btS+lGU+++Tc/-Tmp-/ESS-region/bdd.R at 241':
    >> No such file or directory

    >> [[alternative HTML version deleted]]

    >> ______________________________________________
    >> ESS-help at r-project.org mailing list
    >> https://stat.ethz.ch/mailman/listinfo/ess-help

    > ______________________________________________
    > ESS-help at r-project.org mailing list
    > https://stat.ethz.ch/mailman/listinfo/ess-help