Sun Mon Tue Wed Thu Fri Sat
1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31
|
The Conspiracy against the Human Race
This morning I finished reading the book The
Conspiracy against the Human Race by Thomas Ligotti, who is a contemporary American horror author and reclusive
literary cult figure. I start reading this book on August 20, the day, I
received the book in the mail. Although this book is presented as non-fiction,
while reading, I often wonderd if it is might be just one of his best horror
stories. In fact, (if I remember correctly) Ligotti argues that reality is the greatest
horror imagiable.
The book is inspired by The Last messiah by Peter Wessel Zapffe, a rather obscure, pessimistic Norwegian philosopher,
whoes central thought is that conciousness is a mistake of evolution and that
therefore it would be best if we as humans would stop from getting children.
This all stems from materialism, which states that there is nothing beyond life and that
death is simply the end. There are many good reasons that support materalism,
but one can never be sure that there is indeed nothing outside the universe
we observe.
Because the idea that your existendec ends with death seems to make everything
utterly meaningless (or MALIGNANTLY USELESS, as Ligotti writes), many people
believe that there must be something beyond the observable world.
But if it is indeed the case that everything is just meaningless, and that it
is also impossible to add some meaning to life, it can als be serve as a
form of (stepping stone to) enlightenment because it
leaves the Now as anything there is. Ligotty does not talk about this. From
what he writes (including a quote from
part VII of Confessions by Lev Nikolayevich Tolstoy) one can conclude that he views the 'state' of
enlightmens as but one of the ways to live in denial of reality, to silence
conscious realisation that life ends in death, that all life ends in death,
that our whole universe will one day end of become totally 'frozen'.
I found some parts of the book difficult to read. The forword by
Ray Brassier seems to
be written with the help of a thesaurus to pick the most uncommon words and
expressions.
Today is a palindrome date when
written according to the (D)D-(M)M-(YYY)Y format: 4-10-2014. The
previous such date was October 3, 2013
and the next will be October 5, 2015.
Avoiding merging
I already wrote about merging in distributed version
control and the complex problem of conflict
resolution in my series of postings about concurrency in modeling (as
part of software enginering). When users are changing
a model in close operation, pessimistic concurrency
control it the obvious choice especially if you have many small objects
with relatively little information which are easily created, deleted, split,
joined and/or rearrange in a hierarchy. Many source code management make use of optimistic locking and have tool
support for textual merging edits in source files. But here the number of
objects (files) are small and most edit take place inside the objects
without changing the hierarchy (directory structure).
A disadvantage of pessimistic locking is that it requires access to a
lock server. There is for example a problem with working off-line. This
can be avoided by securing some locks before going off-line and simply
restricting operations that do not match the locks.
In large scale modelling where groups of people are working in seperation on
parts of the model, sometimes without knowing each other, and where parts of
the model are combined in a different location, a centralized repository is
often not the right solution, and pessimistic locking is not possible and
merging seems to be required. But this is not the case. It is possible, and
maybe even desirable, to avoid merging at all. In large scale modelling
changes to parts of the model are not often exchanged but mostly released in
a controlled process often including reviews. In these cases the model is
divided in parts where mostly one group is working at a certain part of the
model at the same time. This means that pessimistic locking can be applied
on the level of parts of the model. But there is a hitch, it is very likely
that there are many reference between objects in the various parts that might
have some consistency requirements. It is important to realize that a
relationship between two objects belong to different parts, must (as an
object) also belong to some part, either one of the two different parts or to
a third part. When a model is divided into parts, one has to decide where
objects that reference objects from different parts should be housed. Dividing
a model into parts along authorities (persons or organisations) who are
responsible for maintaining them, fits very well with the fourth level of
modelling of information systems.
When two parts of a model are released separately, it means that reference
between objects (with a consistency constraint) cannot occur in both
directions. At first this may seem an unwanted restriction, but also from a
semantic point of view it makes sense. If two parts depend on each other (and
changes to them are not released as a unit) it means that when a new release
for the part being referenced is created, the other parts need to be updated
and released again to create a consistent version of the whole model (even if
no real changes to the objects in the part need to be made). It is possible to
define which releases of all the parts form a consistent configuration.
It is very tempting to consider two consecutive releases of a part as a kind
of revisions where objects are simply associated by means of an identifier.
This line of thinking suggest that merging is simple and that global unique
identifier of objects are central. But this is wrong because it is not possible
on the basis of some kind of identifier (name, GUID, URI, IRI, and/or others) to
determine if two objects in different versions are semantically the same,
unless such an association has been formally defined. In effect one need to
define a relationship between the objects of two releases to know how the
changes should be applied in all other parts that reference this part. Such
a releationship could be defined on the basis of a combination of identifiers,
but in many cases, when objects are split and joined between releases, it is
not possible to define a one-to-one relationship.
When creating a new release for a part for a new release of a part it depends
on, it might be needed that for some time, the part contains references to
both releases, to support the process where the references are updated
one-by-one to the object in the new release. In many cases this could be an
automatic process based on the association defined between the two releases,
but for example, when an object is split, it might be that a choice need to
made about to which of the two (or more) objects the reference need to be
made to. This all may seem to be overly complicated, but I think it is best
way to deal with the inherent problems of large scale modelling.
Books
At 12:37, I bought the following two books from bookshop Broekhuis:
- Rekenen op taal by Hugo Battus (Hugo Brandt Corstius),
ISBN:9789021451336, for € 5.00.
- Antiglamour by Carice van Houten and Halina Reijn,
ISBN:9789038898490, for € 8.95.
Amsterdam
I went to Amsterdam to visit the exhibition
'RGB exit' on the last day, because it will be the last opportunity
to see the seventeen works from the serie 'Kleurverhouding' by
Peter Struycken together, because sixteen
of them have been sold and there is a possible buyer for the last one.
I arrived just after half past ten at the central station. From there
I walked to Antiquariaat A. Kok &
Zn on Oude Hoogstraat 14-18. I requested to see a book they had put on
boekwinkeltjes.nl with the title 'P. Struycken' from 1994. When
I saw the book, I immediately told them that it was from 1976.
On my way to bookshop Scheltema, I also
had a look in The American Book Center
and Athenaeum.
At 12:52, I bought the book Susan Sontag: As
Consciousness Is Harnessed to Flesh, Journals & Notebooks 1964-1980
edited by David Rieff, ISBN:9780374100766, for € 6,95.
Shortly after I arrived at Galerie Andriesse•Eyck, Peter Struycken invited me for a drink
and a delicious piece of chocolate cake at the other side of the canal
at Spanjer en
van Twist. The weather was extermely warm for the time of the year
and we could site outside with no coat. Not after long he was called in,
because some guests had arrived. There were quite a number of people
visiting the gallery. I explained the purpose of the painting to some of
them. I also followed some of the discusions Peter had with the visitors.
He is a very kind man with a very broad knowledge, who has really thought
deep about some subjects and has some very interesting ideas to share.
I always enjoy listening to him and see him interact with other people.
In some way, he is thinking like a scientist, and he did write some
scientific publications, at least one journal article but also a book. I
understand he is planning to write another book for which he has gathered
materials in the past years.
Reborn
This morning, I finished reading Susan Sontag:
Reborn, Journals & Notebooks 1947-1963, which I started reading last
Tuesday. The book is edited by David Rieff, the only son of Susan Sontag. It really made me think about some things in my own life.
It is good that it read "Journals & Notebooks" in the title, because
the very varied style of the book. At some times, Sontag gives a very detailed
(hour to hour) description of the events, while at other places months pass
without any note. Het notebooks also contain list of books she want to buy
and/or read. During some periods she only used the notebooks to write down
the many movies (sometimes up to three a day) she watched.
The diamond problem
On wikipedia, the diamond problem is described in the contect of multiple inheritance,
a feature of object-oriented programming languages. But it also occurs in the
case one has versions of document that reference eachother. For example, take
the situation where there is document D that is used as a basis for documents
B and C, and there is also a document A that references those two documents.
Now lets suppose that there is a new version of D, called D'. This possibly
means that documents B and C have to be adapted, if it where only to change
the references in case the changes in D' have no effect. This results in new
versions of documents B and C, called B' and C'. Now as a final step the
document A has to be updated, during which all references to B and C are
changed to B' and C'. (This assumes that all references have an explicite
version indication, and not something like the last available version of B,
B', or any later versions.) In this simple case, when there are only four
documents, it is not difficult to determine when one can start working on
adapting document A, but in case of large scale modelling, there can be
hundreds of versions of documents (or models for that matter) with complex
dependencies, if one wants to enforce the rule that always but one version of
an external document is being(implied) referenced. By this I mean that it
would not be allowed to create a version of document A that references B and
C', because B depends on D and C' depends on D', meaning that A would contain
(implied) references to both D and D'. I am getting the impression that
selecting a set of references that have to be updated simulatiously could even
be a NP-complete
problem. There are also some practical reasons, why one would like to
(temporarily) lift this rule, for example when work on B' has finished but
work on C' is still in progress and might take some time, and one would
already like to start working on incorporating the updates made in B' in
document A. This is another reason why when modeling in the large one should
allow a model to contain references to different versions of an object in
different releases of model that is being referenced as described as a
solution for avoiding merging.
This months interesting links
Home
| September 2014
| November 2014
| Random memories