Engineering legends-flying pig

There is an article on the Air Force's wing website that adds wings to pig. The title party is very serious. In fact, the content is very serious. It is about the inside and outside of the stealth plane. The first generation of the stealth aircraft was F-117's nikhthawk, which was indeed thunderous. It is hard to imagine that something like this can fly and fight. To some extent, it looks really a bit like a pig with wings.
The life of F-117 is also legend. At the time of its development, its confidentiality work had been done by the household, rumors, masks, and smoke scenes. It even produced a toy to deceive the masses and fooled people on Earth for more than a decade. Return to the topic. The F-117 project was around the beginning in 1975 and is also
Masterpiece. After the war, Americans wanted to create a fighter that could escape radar detection. However, due to the current computing capability, the radar reflection characteristics of the curved surface cannot be calculated, not to mention the combination of complex and curved surfaces. If the model is created
Type one by one, pure haystack, and radar wave reflection testing costs a lot, not like the F-104 that with war surplus materials can be flat. As a result, the engineers of Loma have done a lot of geeky design --
Create an airplane using a plane. Since the electromagnetic reflection characteristics of a plane can be computed, splicing a plane of different shapes may create an aircraft that is difficult to detect by radar. Skunk
Works quickly developed a model,
It also verifies its stealth features. Can such an airplane fly? Apparently, it is flying. Not only did it fly, but it also officially went into service this year (the last night e retired in August this year ). Operations in Panama
In the operation, the first bomb was dropped, involved in the Gulf War, air strikes against Iraq, and was shot down by a plane in Serbia. In order not to be detected, F-117 actually paid a huge price. It cannot be supersonic or a drama.
Strong maneuver, no radar, only two 2000lb bombs can be carried. Once detected, it cannot be beaten or run.

However, as the first comprehensive stealth fighter (in fact it is the attacker, take the name of the fighter), F-117 is quite successful. The most critical thing is that it has achieved its fundamental goal-the stealth fighter.
First, it can fly and fly well. Second, to be able to fight, it can throw bombs, although there are only two. Finally, it is invisible and hard to be detected by radar. The Design of F-117 is based on such a core logic: None
How powerful you are. If you can't see your opponent, it will take your life to have a little boy with a Geely blade. As long as you reach the stealth goal, even if you have poor flight performance and a small volume of loaded bombs, it is enough to play a critical role.
Function. As a result, F-117 is used for key tasks targeting critical goals, like cutting your opponent's carotid artery with a razor, a critical blow.
From the engineering point of view, skunk
Works engineers take the initiative to simplify the design and reduce the target requirements. Under the constraints of the current technical conditions, they did not pursue perfect solutions and focused on key features (stealth. In other party
Face, take the initiative to give up secondary goals. This is typical: make it run
First. When designing software, we sometimes fall into the most perfect dilemma. Requires more flexibility, adaptability, scalability, and performance. Eventually, they are lost in the ocean of features. We are often warned, first
Achieve 80% of the features. These 80% features are usually concentrated on 20% features.

Suppose we need to develop a web-based Query Platform. The specific query content and algorithms will change with the customer's changes and requirements, and sometimes there will be on-site rapid implementation requirements. Data also queried
The structure has dynamic characteristics, and some data structures can be determined at runtime. Based on these requirements, we cannot write queries into the code and require scripting. Because it is a Web application
The system can easily implement and change the data obtained from the database and directly convert the data into HTML fragments to insert pages. The best way to implement this task is to use XML
Databases such as DB are directly queried on data using XSLT or XQuery. However, currently
DB is not mature, and its performance is not as good as RDMS. Now we have two options: the well-developed xmldb with excellent performance, or the RDMS is still used, but some auxiliary tools are used. Apparently, the former
Is stupid.

At first, we can manually write SQL statements, organize the result set into a hierarchical structure, extract XML, and then process it with XSLT or XQuery to obtain HTML. Better methods,
Use existing data conversion modules, such as. Net dataset, for result set organization and XML extraction. The things required for this design are at hand and can be quickly implemented. Its shortcomings are:
All the result set data must be extracted before being converted to the complete XML. performance problems may occur when processing large amounts of data. In this step, we can analyze whether further development is required to add features.
First
First, is this solution easy to change. In this solution, an SQL command and An XSLT
Stylesheet. The former is used to extract data from RDMS and organize the data into XML structures. The latter is responsible for constructing HTML and processing some difficult queries in SQL. Both use
Compared with common languages, DSL for data processing makes it easier to execute query tasks. Both of them can be organized in a script file, both of which are text and can be easily modified.
Second, is it possible to quickly implement demand changes on site. Both SQL and stylesheet can be scripted and can be directly modified and run without compilation, and IDE support.
Again, whether the data structure changes during running can be processed. SQL can only rely on a fixed data structure, which is something that has been determined at design time. However, when the result set obtained by the SQL query is converted to XML, it becomes semi-structured data and can have dynamic data structures. XSLT is capable of querying data in this structure.
Most
Then, let's take a look at the performance issues. For general applications, the data volume obtained by the query is not and should not be too large. Tens of thousands of data entries cannot be processed. Therefore, in general
No more than 3 or 5 hundred data records are returned from the next query, which is less than 30 kb. The data volume is converted to XML, and then XSLT conversion is performed.
For systems with 3.0/1 GB of memory, it can be completed within 0.1 seconds. This time is completely acceptable to users. Moreover, when the system runs on a Web server, the execution time is increased by an order of magnitude.
However, after all, many queries need to process a large amount of data. This
Some queries can be further classified. Most of the results are simple result sets, that is, the result set obtained by an SQL query is the results required by the user. These result sets have no hierarchical structure and are completely
Table. For such a query, You can first segment the data, for example, 100 pieces of data, convert a piece of data into XML, then execute XSLT, send the result to the client for display, and then process the next section. This
The user does not have to wait until all the data is extracted and processed. The user can see the data almost immediately, and the subsequent data will be processed by the user's time to read the first batch of data. This segment processing mechanism can easily
Joining the Query Platform does not require a lot of work.
At this point, most of the query requirements (about 80%) are covered. For those data with a large volume of data that has a hierarchical structure, data that cannot be processed in segments, or data that cannot be processed in fixed segments, do not consider it. Once required, you can customize a dedicated module for processing.

F-117 is not the end of the story of fliggy, but the beginning. With the development of technology, engineers have the computing power to process simple surfaces. So, immediately applied it to the plane, the birth of B-2 spirit
Born. In any case, the B-2 looks more like a normal plane, even though it is not a conventional plane. Only from the data point of view, the B-2 is powerful, the maximum takeoff weight of 170 tons, Range
11000 km, with a bomb volume of 22.7 tons, can carry almost all of the US military active bombs. The latest improvement allows it to drop 80 bombs at the same time, hitting 80 different targets. Compared with f-
117, which is already a qualitative leap. Although its body is huge, its stealth capability is more advanced than F-117. This is the inevitable result of key technological breakthroughs. However, B-2 is not enough, in the late 1980s s, the development of IT technology makes people have the ability to process complex surface computing. As a result, Americans spared no effort to develop the ultimate aircraft-F-22 Raptor. The ability of F-22 can be illustrated with only one digit: In the F-22vsF-15/16 exercise organized by the US military, the F-22 has achieved114:1.
This
The success story of three generations of Stealth Fighters once again emphasizes a classic practice in Engineering: make it run
First. In the initial stage, efforts were made to achieve the most critical and core functions, which were sufficient to meet most application needs. Other demands are temporarily avoided or implemented through traditional means. With
With the development and progress of technology, the technical conditions of those peripheral functions gradually mature, and the corresponding functions will naturally be implemented.

With the large-scale application of XML and Its Related Technologies, various traditional RDMS vendors gradually introduce XML-related technologies in their databases. For example, XML/SQL,
XML clause. These technologies allow us to directly construct a result set into a hierarchical XML document in SQL statements. Using these technologies, a large amount of data has a hierarchical result set
The results set of fixed segments can be segmented. With these technologies, queries that cannot be implemented in the query system become scarce. At this time, we can also think that this query
The system can meet all the applications that face it, without having to spend more energy to meet the needs of the remaining Millennium.

However, when using this engineering method, there is a prerequisite that key target features must be correctly identified. If we use the big data processing capability as a key target in the case of a query system
Obviously, this design will not be used to deviate from the real goal. For the identification of key target features, you can examine the needs, find the features that meet the most needs, and sort and analyze them.