W3C home > Mailing lists > Public > public-xml-processing-model-wg@w3.org > April 2006

Re: Side effects and interoperability

From: Innovimax SARL <innovimax@gmail.com>
Date: Wed, 19 Apr 2006 16:02:36 +0000
Message-ID: <546c6c1c0604190902l191c008aoea31293eb7fdf443@mail.gmail.com>
To: "Jeni Tennison" <jeni@jenitennison.com>
Cc: public-xml-processing-model-wg@w3.org


On 4/13/06, Jeni Tennison <jeni@jenitennison.com> wrote:
> Hi,
> [Having read the minutes of the telcon.]
> I don't think that no side effects impacts on whether caching is allowed
> or not. A component with side effects can consistently produce the same
> side effects each time, after all. Rather, whether or not we allow side
> effects is important when considering reordering steps, or skipping them
> altogether.


> I'm in favour of a "no side effects" rule because I want to make sure
> that the same pipeline run at the same time on different pipeline
> engines produces the same result.

Not sure, I think we use "side effetct less" for "context independant"

if an xslt instance do into it's output document :
- inserting current time and date (non cachable)
- format some data depending on localization (timezone, on the
currency, the language etc....) (non repeatable)
- insert information about the processor (non indenpence on the vendor)

it could be considered as side effect less (no data have been altered
other than the output) but couldn't be considered to be reproduced
I think we should be more precise about this point

> I view that as the basis for
> interoperability, which is one of our design principles.
> I also want pipeline engines to be able to optimise, such as by not
> performing steps that don't need to be performed or by performing steps
> in parallel. I don't want those optimisations to effect the final result.

Agree totally, but we should inforce the rule to respect so that it
could be the case.

> I therefore think it's vital that if the pipeline language doesn't
> constrain a pipeline engine to run two steps in a particular order then
> the order in which the two steps are run doesn't have an effect: the
> steps have no side effects.

again agree, but with changing "side effect" (see above)




Innovimax SARL
Consulting, Training & XML Development
9, impasse des Orteaux
75020 Paris
Tel : +33 8 72 475787
Fax : +33 1 4356 1746
RCS Paris 488.018.631
SARL au capital de 10.000 
Received on Wednesday, 19 April 2006 16:06:02 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:32:39 UTC