Re: matlab speed issues



On 16 Sep, 14:23, "Lee Baker" <eelrekab+mat...@xxxxxxxxx> wrote:
Hi Rune,
thanks for your reply. To give you some more information about the nature of my problem:

1. Which is fastest - publishing a 'new' page or an existing one (i.e.. should I create original published pages or overwrite existing ones)?

Wrong question. The key is not how fast matlab produces
a page, but how easy it is for you as user to find the
correct page. If there are several versions of the same
page available, you (or some other person) will become
confused as to which one is updated and which ones are
obsolete. Such confusion will waste far more time down
the line than you might save on a file write operation.

What my program does is 'publish' a page (for example, 'page0'), and since it iteratively performs the same analyses for different variables, 'page0' needs to be iteratively renamed as 'page1', 'page2', etc..
My question is whether 'publish' is quicker when writing a page that does not already exist (i.e. 'page0' is written and then renamed as 'page1' by the 'movefile' function) or when writing a page that already exists (i.e. 'page0' - which already exists - is overwritten by 'publish' and then copied to 'page1' by the 'copyfile' function).

Maybe you can save some time by using style sheets that specify
the whole appearance of the pages, only write the case-specifics
to a data file or figure, and then write a small file that
presents the data and specify the style sheet.

I can't really see that writing a couple of files should take that
long, *assuming* that alle the computations have been done up front.
Could it be that the time for the actual computations are included
in the call to 'publish'?

Rune
.