On Fri, Sep 21, 2018 at 12:44 PM Hans Hagen <j.hagen@xs4all.nl> wrote:
On 9/21/2018 12:05 PM, luigi scarso wrote:
>
>
> On Fri, Sep 21, 2018 at 11:06 AM Taco Hoekwater <taco@elvenkind.com
> <mailto:taco@elvenkind.com>> wrote:
>
>     Hi,
>
>      > On 21 Sep 2018, at 10:59, luigi scarso <luigi.scarso@gmail.com
>     <mailto:luigi.scarso@gmail.com>> wrote:
>      >
>      >
>      > % start demo
>      > \enabledirectives[system.callbacks.permitoverloads]
>      > % previous line allows redefinition of ‘wrapup_run’
>      > iirc ,
>      > this should be avoided ….
>
>     Sure, but I find wrapup_run extremely useful, and I know of no other
>     way to use it.
>
>
> sure but the idea of modifying the context state (broadly speaking)
> before that context says that the state is consistent
> (ie the run is  finished, in this case) is , how to say.. hm hm.
> Maybe signalling with an asyn msg that has nothing todo with the context
> state (just to say "Hey, context here: I am finishing the run" to
> somebofy else) ,
> but usually one wants  also to check the starts and the end of the run
> as process, ie just before the run starts and just after the run ended.
> This case seems safe but I have already dubious on \
> $ lua -e ' os.execute("ls \c*") '
> lua: (command line):1: invalid escape sequence near '\c'
> while
> $ lua -e ' os.execute([[ls \c*]])'
> is ok
> Also,  I run context foo.tex and at the end  of the runs I found foo.pdf
> (expected) and doc-<idcode>.pdf ("hm .. where does it come from ? I dont
> remember ...I have to look into the source" )
> so two times the space --- and the copy is done at each run, iirc.
>
>
> But it's just to complete the picture:  once one  knows the limits, it's
> a useful callback.
that callback is really the last ... files have been closed then
sure , but after it there are 2 free_<something> and free the Lua state, 
--- and closing a file can be delayed by the OS. Anyway, this case looks  ok.
 
--
luigi