From: Robert Ramey (ramey_at_[hidden])
Date: 2002-09-10 22:29:28
On Tuesday, Sep 10, 2002, at 03:33 US/Pacific, Vladimir Prus wrote:
>> Robert Ramey wrote:
>>> Fellow boosters,
>>> I am loading my final offering for serialization.zip and requestion a
>>> formal review.
>>> This version differes from the previous one in the following:
>>> 1) Additions to documentation to explicitly address issues of
>>> exception safety.
>>> 2) More test cases/demos to illustrate handlling of the above issues.
>>> 3) Additions to documentation to include rationale for not depending
>>> on type_id
>> I'm afraid I'm not convinced by either the rationale or your posts.
>> For that reason plan to argue and vote against the inclusion of your
>> library in the current state.
I'm not sure which posts or rationale you are unconvinced by. I can say
I tried to address all points raised only after careful study and consideration
and I am convinced that they have been addressed in the best way. My
reasons for the decisions made have been documented in posts and the
rationale. In the absence of new facts or observations, I see no reason to
make any more changes.
>> In addition, there was *no* real discussion on interface. I still
>> would like to use 'describe', which may work in 95% of cases. I also
>> have some other minor disagreements. I'm afraid we won't be able to
>> come to a decision over formal review period.
There was quite a bit of discussion on the interface - particularly regarding
the describe functionality. I had nothing against it in priniciple, its just
that in practice it didn't really add anything.
>> Given that a lot of time still remains, would you be willings to
>> actively discuss those issues. I really would like to have
>> serialization in Boost, only don't feel that the library reached the
>> level that I'd like.
I've been answering issues raised since april. There has been 5 months
and 4 drafts. This is ample opportunity to raise new issues. All issues
raised durring this time were given much consideration. The library
has surpassed its original list of objectives by a significant margin. Its
been months since new issues have been raised.
On Tuesday, Sep 10, 2002, at 03:33 US/Pacific, Matthias Troyer wrote:
>Robert, I appreciate all your efforts in writing this library.
>However, while different formats are now supported, I mentioned
>some time ago that a specialization for writing C-style arrays of
>the basic data types is important for good performance. When
>(de)serializing a matrix of 100'000'000 doubles I do not want to
>call (in the worst case) a virtual function 100'000'00 times.
This issue is explictly addressed by a section of the reference manual under
the section of serialization of large numbers of small objects. I feel that
your concerns are more than adequatly addressed here.
>And note, I still want to serialize these numbers in a portable
>format and not just write the machine-dependent binary representation.
Rather than include every desired format - a hopless quest, the library format
is decoupled from the serialization itself. Included are three formats - ascii and
wide character text as well as native binary. These demonstrate how simple
it is to use ones own preferred format without changing any of the code in the
library itself. Thus, dt is quite simple to specify ones own format (e.g. XDR)
without any changes to the library. I would hope
that would address everyones need for their favorite format.
>I would thus also prefer some more discussions before a formal review.
I see no issues new issues raised here. And certainly no aspect of the original
list of objectives has been overlooked. Also, I know that a number of people
have been using the draft libraries with good success.
So as it stands, I have nothing to add - its time to vote.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk