On Sun, 28 Jul 2013, John E. Malmberg wrote:

>> I'm fine with this approach. You call it a "draft patch", does it mean
>> you'll provide another one or are you fine with this for now?
>
> I would prefer if there was someway to cache the data from the first stat()
> call so that the second stat() call later was not needed, as stat() has
> higher overhead on VMS than on Unix.

Ah, ok. Let's consider that an optimization then and we can work on that in a
separate stage.

For a solution, I can imagine storing the stat struct results in the FormData
struct in the first phase and then read from there in the second phase
instead of the second stat().

> I see one other file open for read in libcurl, but have not had time to look
> at it to determine the impact on VMS, or a reproducer. It should have the
> same issues as with this one and FTP upload patch committed earlier, so a
> similar solution should be needed.

I see. Still a separate issue so we can take that in a separate change!

Thanks for all your hard work, I'll merge this one now!

-- 
  / daniel.haxx.se
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette:  http://curl.haxx.se/mail/etiquette.html