<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Programmers Don’t Read Books - Medium]]></title>
        <description><![CDATA[But You Should!  —Reading  Notes on Programming/Computer Science/Computer Engineering books - Medium]]></description>
        <link>https://medium.com/programmers-dont-read-books?source=rss----d2f7441a9429---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Programmers Don’t Read Books - Medium</title>
            <link>https://medium.com/programmers-dont-read-books?source=rss----d2f7441a9429---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 13:30:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/programmers-dont-read-books" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Information Retrieval 1]]></title>
            <link>https://medium.com/programmers-dont-read-books/information-retrieval-1-e81273c28a28?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/e81273c28a28</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Mon, 02 Mar 2015 00:48:29 GMT</pubDate>
            <atom:updated>2015-03-02T00:48:19.559Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/782/1*l8mMnyrbeHAdyKYAlftqbA.gif" /></figure><p>Overview</p><p>1. Information makes users overwhelmed</p><p>2. Process of finding materials that satisfied an information need</p><p>3. Does not answer a specific question; retrieves the doc that could be used to answer that question</p><p>4.</p><p>a. retrieving <strong><em>relevant</em></strong> doc to a query</p><p>b. large set efficiently</p><p>5. Earlier literature, information seeking as a form of problem sovling method</p><p>6. information need is a pyramid; only its peek is made visible by users</p><p>7. semantic aspect 语义方面 meaning of word; syntactic aspect句法方面 order of word; interaction and feedback; authority and authenticity</p><p>8. History</p><p>9. IR vs DR</p><p>10. information storage and retrieval — aboutness</p><p>11. traditional IR system;</p><ul><li>get a set of keywords</li><li>stopwords eliminated; words are stemmed</li><li>user interface</li><li>enhanced retrieval</li></ul><p>12. enhanced retrieval</p><ul><li>auto categorization</li><li>spam filtering</li><li>info routing — ?</li><li>auto clustering</li><li>recommending</li><li>info extraction</li><li>info integration</li><li>question answering</li></ul><p>12. type of IR system</p><p>13. keyword search</p><p>problems:</p><ul><li>language ambiguities</li><li>spamdexing — strange web page full of keywords</li></ul><p>14. boolean search</p><ul><li>and / or are not perfect</li><li>usually very precise</li><li>non-ranking-able</li></ul><p>15. vector space model</p><ul><li>document indexing — keywords</li><li>term weighting</li><li>similarity ranking of doc — cosine coefficient</li><li>latent semantic analysis</li><li>probabilistic models — subjective; independent</li></ul><p>16. language models</p><ul><li>NLP</li><li>lexical ambiguity</li><li>inefficiency</li></ul><p>17. properties of IR system</p><p>18. various hypothesis — relevance hypothesis; associated hypothesis; cluster hypothesis</p><p>19. word stemming</p><p>20. ranked retrieval — 需要精辟</p><p>21. Challenges for IR systems</p><ul><li>language and information need complexities</li><li>relevance complexities</li><li>word mismatch complexities</li><li>modern spelling complexities</li><li>paraphrasing complexities</li><li>anaphora complexities — it; them…</li></ul><p><strong><em>22. challenges associated with individual differences in IR — could be my paper’s topic</em></strong></p><ul><li>training both user and the system</li></ul><p>23. use semi-structured data instead of database — database more for structured data</p><p>24. concept-based IR</p><ul><li>keyword-based IR model — commonly used in search engines</li><li>concept-based IR model — 我曾经尝试找对于该关键字的描述，或者经常co-occur的keywords，把这些当做是concept的word bag</li><li>semantic linguistic network of concepts — Nowadays’ google</li><li>Thesaurus — manual work; synonymy; antonymy</li><li>predictive model — neural networks;</li></ul><p>25. features of conceptual structures</p><ul><li>type of conceptual structure</li><li>form of representation — tree; semantic network; context vectors;</li><li>relationship supported by a conceptual structure — subsumption; a kind-of; a part of; association; and / x-or</li><li>creation of conceptual structure — manual creation; automatic learning; NLP</li></ul><p>26. IR and related areas</p><ul><li>focus on user aspect</li><li>AI — first order predicate logic and bayesian networks</li><li>NLP</li><li>Machine learning</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e81273c28a28" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/information-retrieval-1-e81273c28a28">Information Retrieval 1</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Linux Device Drivers Ch01]]></title>
            <link>https://medium.com/programmers-dont-read-books/linux-device-drivers-ch01-9a4091e55e1b?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/9a4091e55e1b</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Sun, 07 Dec 2014 11:15:49 GMT</pubDate>
            <atom:updated>2014-12-07T11:15:34.424Z</atom:updated>
            <content:encoded><![CDATA[<p>First glance</p><p>1. The distinction between mechanism and policy is one of the best ideas behind the Unix design</p><ul><li>“what capabilities are to be provided” (the mechanism)</li><li>“how those capabilities can be used” (the policy)</li></ul><p>2. The role of the device driver</p><ul><li>some basic concept</li><li>be as policy free as possible</li></ul><p>3. Splitting the kernel</p><ul><li>process management; memory management; file systems; device control; networking</li></ul><p>4. Loadable Modules</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/787/1*l835gCQN7lJXTTeSEhOcpw.png" /></figure><p>5. Classes of Devices and Modules</p><ul><li>decomposition is a key element of scalability and extendability</li><li>so create a different module for each new functionality</li><li>Character Devices — can be accessed as a stream of bytes</li><li>Block devices — a block of device that can host a filesystem; like /dev</li><li>Network interfaces — Not being a stream-oriented device; hard to be mapped to a node in the file system</li></ul><p>6. Security Issues</p><p>7. Version numbering — stable and unstable</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9a4091e55e1b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/linux-device-drivers-ch01-9a4091e55e1b">Linux Device Drivers Ch01</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Computer Networking — A Top-down Approach Chpt 5]]></title>
            <link>https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-5-f2e3680f7d74?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/f2e3680f7d74</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Wed, 19 Nov 2014 08:45:28 GMT</pubDate>
            <atom:updated>2014-11-19T08:45:17.861Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*ZtPVVZ258uktzIVdwGgnWg.gif" /></figure><blockquote><em>CHAPTER 5</em></blockquote><p>1. the transportation mode is a link-layer protocol, and the travel agent is a routing protocol</p><p>2. Framing; Link access; Reliable delivery; Error Detection and Correction</p><p>3. Implemented in a network adapter; Network Interface Card — much of link layer services are implemented in hardware</p><p>4. the link layer is a combination of hardware and software</p><p>5. error-detection and -correction — 2 dimensional parity scheme can be use to detect and correct</p><p>6. Checksum — TCP/UDP check all header and data; IP only check IP header; checksum is designed to be fast and easy for software implementation</p><p>7. Cyclic Redundancy Check (CRC) — polynomial codes; hardware; strange calculation</p><p>8. Multiple Access Protocol — channel partitioning protocols; random access protocols; and taking-turns protocols</p><p>9. Channel Partitioning protocols — use TDM, time-division multiplexing; each node has rate of R/N; must wait for its turn; use FDM; speed is limited</p><ul><li>code division multiple access: CDMA; also a channel partitioning protocol</li></ul><p>10. Random Access Protocol — delay randomly before transmitting</p><ul><li>slotted ALOHA — transmit at the beginning of slots; p probability to transmit at current new slot</li><li>ALOHA — no fixed slot; collision occurs during transmitting, finish and retransmit with probability p</li><li>Carrier Sense Multiple Access (CSMA) — listen before speaking; if speak at same time, then stop; CSMA and CSMA/CD (with collision detection); channel propagation delay; CSMA/CD contains binary exponential backoff algorithm; new frames are allowed to start their own detect algo; efficiency equition</li></ul><p>11. Taking-turns Protocols</p><ul><li>master node polls each of node in a round-robin fashion; polling protocol; eliminates the collisions and empty slots;</li><li>token-passing protocol — who hold the token can send frames; after sending, send token to next node</li></ul><p>12. DOCSIS: The Link-Layer Protocol for Cable Internet Access — MAP frames for downsteam; minislot request frames in upstream</p><p>13. Link-layer Addressing and ARP</p><p>14. MAC address — physical address; 6 byte long; broadcast address all 1</p><p>15. ARP — Address Resolution Protocol; Used to get MAC address; Each host and router has an ARP table; Flow: construct ARP packet, broadcast, receive from unicast;</p><blockquote>A hub is a physical-layer device that acts on individual bits rather than frames. When a bit, representing a zero or a one, arrives from one interface, the hub simply re-creates the bit, boosts its energy strength, and transmits the bit onto all the other interfaces.</blockquote><p>hub-based star topology</p><p>16. Ethernet Frame Structure</p><ul><li>Data field — network packets/datagram</li><li>Destination / Source address</li><li>Type field — ARP 0x0806;</li><li>Cyclic redundancy check (CRC)</li><li>Preamble — Each of the first 7 bytes of the preamble has a value of 10101010, Last 10101011 → 2 consecutive 1s alert adapter at source: data is about to come</li></ul><p>17. Ethernet simple and cheap — connectionless; no negetive ack; no shake hands</p><p>18. Switches</p><ul><li>Forwarding and Filtering — switch table; depends on interface &amp; mac address</li><li>Self-learning — plug-and-play devices</li></ul><p>19. Properties of link-layer switching — Elimination of collisions; Heterogeneous links, can hold different devices with different media / speed; Mangement</p><p>20. Switches vs Routers</p><p>21. virtual local area networks (VLANs)</p><p>22. Link Virtualization: A Network as a Link Layer //TODO</p><p>23. Data Center Networking</p><p>24. <strong><em>Retrospective</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f2e3680f7d74" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-5-f2e3680f7d74">Computer Networking — A Top-down Approach Chpt 5</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Computer Networking — A Top-down Approach Chpt 4]]></title>
            <link>https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-4-18f7c6e438ff?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/18f7c6e438ff</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Sun, 09 Nov 2014 06:51:57 GMT</pubDate>
            <atom:updated>2014-11-09T06:51:57.880Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2ps4KlEODqLbgYt9u5Li7w.png" /></figure><h3>Computer Networking — A Top-down Approach Chpt 4</h3><blockquote><em>CHAPTER 4</em></blockquote><p>1. Forwarding &amp; Routing — difference; forwarding table</p><p>2. communication setup — require the routers along the chosen path from source to destination to handshake</p><p>3. Network Service Models — Constant bit rate (CBR) ATM network service; Available bit rate (ABR) ATM network service</p><p>4. virtual circuits (VCs)</p><blockquote>Computer networks that provide only a <strong>connection </strong>service at the network layer are called virtual-circuit (VC) networks</blockquote><blockquote>computer networks that provide only a <strong>connectionless</strong> service at the network layer are called datagram networks</blockquote><p>5. VC setup</p><blockquote>routers along the path between the two end systems are involved in VC setup, and each router is fully aware of all the VCs passing through it.</blockquote><p>6. Datagram Networks — VC network is arguably more complex than a datagram network</p><blockquote>7. Router</blockquote><p>Input ports; Switching fabric; Output ports; Routing processor</p><p>8. Input port processing — termination function →link-layer processing →lookup, forwarding queue →switch fabric</p><p>9. Switch — Heart of a router; switch is forwarding; via memory; via a bus; via interconnection network</p><p>10. Output processing — packets may lost during queue; buffer size should be very large to suit the queuing</p><p>11. <strong>Packet scheduling</strong> plays a crucial role in providing <strong>quality-of-service guarantees</strong></p><p>12. Random Early Detection (RED) algorithm — when buffer full</p><p>13. head-of-the-line (HOL) blocking will happen</p><blockquote>14. IP: forwarding and addressing in the internet</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/933/1*vS-4nG0THRwor8cVpl31ag.png" /></figure><p>15. Network-layer packet is referred to as a <strong><em>datagram</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/810/1*vNP6PVova43lcz_53R5EAA.png" /></figure><p>16. Raison d’être — reason for existence</p><p>17. The maximum amount of data that a link-layer frame can carry is called the <strong>maximum transmission unit (MTU)</strong></p><p>18. if IP datagram &gt; MTU → fragment the data into smaller IP datagrams</p><blockquote>19. an IP address is technically associated with an interface, rather than with the host or router containing that interface.</blockquote><p>20. 192.168.10.10 is a <strong>dotted-decimal notation</strong></p><p>21. 223.1.1.0/24; /24 is a subnet mask</p><p>22. The Internet’s address assignment strategy is known as <strong>Classless Interdomain Routing (CIDR </strong>— pronounced cider<strong>)</strong></p><blockquote>23. use longest prefix matching to route toward small subnet</blockquote><p>24. 255.255.255.255 broadcast address</p><p>25. ISP &amp; its customers’ subnet</p><p>26. Obtaining a Host Address: the Dynamic Host Configuration Protocol</p><p>27. DHCP server discovery use DHCP discover message — destination broadcast IP address, source IP; DHCP server offers — may get several; DHCP request; DHCP ACK</p><p>28. Network Address Translation (NAT) — NAT translation table is to know the internal host； NAT router will change IP address &amp; port number for source</p><p>29. Peer behind NAT cannot directly contact another peer; connection reversal is needed; NAT traversal</p><p>30. UPnP — Universal Plug and Play; ask NAT to create a hole that map (private IP, port) → (public IP, port)</p><blockquote>31. Internet Control Message Protocol (ICMP)</blockquote><p>32. network layer of the internet has 3 main components: IP protocol; internet routing protocol; ICMP</p><p>33. ICMP lies architecturally above IP — ICMP msg is inside of IP datagrams; first 8 bytes of the IP datagram that caused error; Traceroute is implemented with ICMP messages;</p><p>34. use traceroute [-T][-I]; mtr google.com.sg --report; tracepath</p><p>35. IP v6</p><p>every grain of sand on the planet can be IP-addressable;</p><p>anycast address — deliver datagrams to any one of a group hosts;</p><p>40-byte header;</p><p>Flow labeling and priority — it is a foresee; 20-bit;</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/843/1*xcZatbE7625lTJROG7fjGg.png" /></figure><p>36. Transitioning from IPv4 to IPv6 — IPv6-capable systems can be made backward-capable; dual-stack approach;</p><p>37. tunneling — IPv4 encapsulating IPv6</p><p>38. IP Security — IPsec is connection-oriented;</p><blockquote>39. Routing Algorithms</blockquote><p>40. goal: find a shortest path</p><p>41. Link-State (LS) Routing Algorithm — Dijkstra’s algorithm</p><p>42. The Distance-Vector (DV) Routing Algorithm — bellman-ford algorithm; decentralized</p><blockquote>DV-like algorithms are used in many routing protocols in practice, including the Internet’s RIP and BGP, ISO IDRP, Novell IPX, and the original ARPAnet.</blockquote><p>43. Routing loop can happen when y think z is shorter way, z think y is shorter way</p><p>44. use poisoned reverse to lie</p><p>45. Hierarchical Routing — Scale; Administrative autonomy; Autonomous System; add gateway routers to response outside</p><p>46. inter-SA &amp; intra-SA</p><blockquote>47. Routing the Internet</blockquote><blockquote>An <strong>intra-AS</strong> routing protocol is used to determine how routing is performed within an autonomous system (AS). Intra-AS routing protocols are also known as <strong><em>interior gateway protocols</em></strong></blockquote><p>48. Intra-AS Routing Information Protocol — RIP uses the term <strong>hop</strong>, which is the number of subnets traversed along the shortest path from source router to destination subnet, including the destination subnet; RIP response message/30sec;</p><p>49. Intra-AS Routing in the Internet: OSPF — OSPF is a <strong>link-state protocol</strong> that uses flooding of link-state information and a <strong>Dijkstra</strong> least-cost path algorithm; Can be configured hierarchically;</p><p>50. Inter-AS Routing: BGP — Border Gateway Protocol; TCP connection to send BGP msg is <strong>BGP session</strong>; external BGP (eBGP) session span 2 ASs; internal BGP (iBGP) session in the same AS;</p><p>51. //TODO BGP read again</p><p>52. Inter-AS vs Intra-AS — differences between the goals: Inter-AS Policy-oriented; Inter-AS for scale; Intra-AS focuses more on performances</p><p>53. <em>de facto —</em> the truth/fact</p><blockquote>54. Broadcast and Multicast Routing</blockquote><p>55. Broadcast Routing Algorithms — N-way-unicast: duplicated packets from source; Uncontrolled flooding; Controlled flooding — add source ip | if not shortest path to me, i will not send out; Spanning-Tree Broadcast — centre-based spanning tree;</p><p>56. Broadcast Algorithm in Practice</p><blockquote>Broadcast protocols are used in practice at both the application and network layers.</blockquote><p>57. Multicast — how to identify the receivers of a multicast packet; how to address a packet sent to these receivers</p><p>58. Internet Group Management Protocol — join/leave a multicast group</p><p>59. Multicast Routing in the Internet — Distance-vector Multicast Routing Protocol;</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=18f7c6e438ff" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-4-18f7c6e438ff">Computer Networking — A Top-down Approach Chpt 4</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Computer Networking — A Top-down Approach Chpt 3]]></title>
            <link>https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-3-71f774d31cfd?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/71f774d31cfd</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Sun, 02 Nov 2014 07:56:36 GMT</pubDate>
            <atom:updated>2014-11-29T09:41:16.710Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*UjNfQ2QR1lysl9ikYsBOew.jpeg" /><figcaption>* 3 by visualizing</figcaption></figure><h3>Computer Networking — A Top-down Approach Chpt 3</h3><blockquote><em>CHAPTER 3</em></blockquote><p>1. Transport-layer protocol provides for logical communication between application processes running on different hosts</p><p>2. compared to network layer — which is the logic communication between hosts</p><p>3. TCP — reliable data transfer; congestion control</p><p>4. Multiplexing &amp; demultiplexing — transport to network &amp; network to transport</p><p>5. port number — 0 ~ 65536; 0 ~ 1023 well know ports</p><p>6. connectionless transport — UDP; DNS service; RIP routing table updates</p><p>7. UDP segment structure — UDP checksum</p><p>8. Data transfer protocol — FSM</p><p>9. ACK &amp; NAK — need putting sequence number</p><p>10. Stop-and-wait versus pipelined protocol — most of the time, sender is waiting. Because, sender sent the package out very fast. The most time consumption is the RTT. So we should use pipeline</p><p>11. Go-Back-N</p><blockquote>N is often referred to as the window size and the GBN protocol itself as a sliding-window protocol.</blockquote><p>12. Selective Repeat</p><blockquote>As the name suggests, selective-repeat protocols avoid unnecessary retransmissions by having the sender retransmit only those packets that it suspects were received in error (that is, were lost or corrupted) at the receiver</blockquote><p>13. Now comes to TCP; just now was general principles of reliable data transfer</p><blockquote>14. TCP</blockquote><p>Required knowledge: error detection, retransmissions, cumulative acknowledgments, timers and header fields for sequence and acknowledgment numbers.</p><p>TCP — connection-oriented, full-duplex service, always point-to-point, three-way handshake</p><p>TCP buffer — maximum segment size MSS or maximum transmission unit MTU</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/724/1*Ri10ZqFYM7s8RLHc9TY4Ew.png" /></figure><p>TCP Sequence number &amp; acknowledgement number — Seq #= previous_it + MSS; Ack # = receive_id + 1 → expecting</p><p>15. RTT — use Exponential Weighted Moving Average (EWMA)</p><p>EstimatedRTT = (1 – α) * EstimatedRTT + α <strong>* </strong>SampleRTT</p><p>SampleRTT = currentRTT; changes overtime;</p><p>EstimatedRTT = smoother RTT; used to set timeout period</p><p>DevRTT is EWMA of difference between SampleRTT &amp; EstimatedRTT</p><p>DevRTT = (1 – β) * DevRTT + β * | SampleRTT – EstimatedRTT |</p><p>16. TCP timeout</p><p>TimeoutInterval = EstimatedRTT + 4 * DevRTT</p><p>Timeout and Retransimission</p><p>17. Doubling Timeout — be polite</p><p>18. Fast Retransimit — duplicate ACK</p><p>19. TCP flow-control — 1 receive windows + 1 tcp application buffer = RcvBuffer</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/912/1*HaykHYWBxYSbWjgx54cyOg.png" /></figure><blockquote>LastByteRead: the number of the last byte in the data stream read from the<br>buffer by the application process in B</blockquote><blockquote>LastByteRcvd: the number of the last byte in the data stream that has arrived from the network and has been placed in the receive buffer at B</blockquote><p>For sender: LastByteSent – LastByteAcked &lt;= rwnd</p><p>rwnd is the “windows size” in TCP header, which indicate how many data can be receive</p><p>20. TCP connection management — P253 →step 123, need to set ACK flag to 1 or not?</p><p>21. FIN flag is set to tear down the connection</p><p>22. SYN flood attack</p><p>23. Congestion Control — available bit-rate (ABR); asynchronous transfer mode (ATM)</p><p>a. large queuing delay</p><p>b. packets lost due to route buffer overflow</p><p>c. unneeded retransimission</p><p>d. due to congestion</p><p>24. Approaches to congestion control — end2end control, tcp must use; network-assisted congestion control, routers feedback to senders, network layer</p><p>25. resource-management cells — convey congestion-related information among the hosts and switches</p><p>26. TCP congestion control — end-to-end congestion control</p><p>27. LastByteSent – LastByteAcked &lt;= min{cwnd, rwnd}</p><p>28. TCP self-clocking</p><p>29. slow start — exponential growth; ssthresh, slow start treshold;</p><blockquote>when the value of cwnd equals ssthresh, slow start ends and TCP transitions into congestion avoidance mode</blockquote><p>30. Congestion Avoidance — Tahoe vs Reno</p><p>31. TCP fairness</p><blockquote>In particular, it has been shown that when multiple connections share a common bottleneck, those sessions with a smaller RTT are able to grab the available bandwidth at that link more quickly as it becomes free (that is, open their congestion windows faster) and thus will enjoy higher throughput than those connections with larger RTTs</blockquote><p>32. Fariness &amp; UDP — UDP are always not friendly</p><p>33. Use parallel TCP connection to dominate network</p><p>Update:</p><p>34. ssthresh is initialized to a large value; 64kB</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=71f774d31cfd" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-3-71f774d31cfd">Computer Networking — A Top-down Approach Chpt 3</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Computer Networking — A Top-Down Approach Chpt 2]]></title>
            <link>https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-2-e00a32b6882f?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/e00a32b6882f</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Thu, 23 Oct 2014 13:58:32 GMT</pubDate>
            <atom:updated>2014-10-25T03:29:53.728Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/425/1*j3fd34IEJvtpF5mhVEksjQ.jpeg" /></figure><h3>Computer Networking — A Top-Down Approach Chpt 2</h3><blockquote><em>CHAPTER 2</em></blockquote><p>1. Network Application — Server-Client; P2P</p><p>2. Socket — interface between process</p><p>3. Transport Services: Throughput, timing, security</p><p>4. SSL — Secure Socket Layer</p><p>5. TCP vsUDP</p><p>6. Application Layer Protocals</p><p>7. Web &amp; HTTP</p><p>8. Non-Persistent and Persistent Connections</p><p>9. Three-way handshake</p><p>10. Round-trip time — RTT; Small package to travel from client to server</p><p>11. HTTP Msg &amp; Response Format</p><p>12. Cookie — a nice Case study</p><p>13. Web Caching — Proxy Server; installed by ISP; Maybe university or major residential; calculation is similiar to cache</p><p>14. Content Distribution Networks (CDNs) — shared CDNs; dedicated CDNs (Google)</p><p>15. FTP — Similiar to HTTP, but server needs to maintain state about the user</p><p>16. SMTP</p><p>17. SMTP →POP3; IMAP—Post Office Protocol V; Internet Mail Access Protocol</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wbtsiAqzM4sKbS_86k3K_w.png" /></figure><p>18. DNS — services: Host aliasing; Mail server aliasing; Load Distribution; LOAD Distribution</p><p>19. DNS →a wonderful example of distributed database. problems: A single point failure; Traffic volume; Distant centralized database; Maintenance</p><p>20. Root DNS servers; Top-level domain (TLD) servers; Authoritative DNS servers; local DNS server</p><blockquote>The query from the requesting host to the <strong>local DNS server</strong> is recur-<br>sive, and the remaining queries are iterative.</blockquote><p>21. DNS Caching</p><p>22. DNS Record type / Format</p><p>23. p2p applications — //TODO here P145 ~ P156</p><p>24. Socket Programming — UDP, TCP</p><p>25. Port number</p><blockquote>attaching the source address to the packet is typically not done by the UDP application code; instead it is automatically done by the underlying operating system.</blockquote><p>26. Python code = =</p><p>27. TCP — connection-oriented; need handshake; the connection is fixed and can be used directly</p><blockquote>During the three-way handshake, the client process knocks on the welcoming door of the server process. When the server “hears” the knocking, it creates a new door — more precisely, a new socket that is dedicated to that particular client.</blockquote><blockquote>With TCP, all bytes sent from one side not are not only guaranteed to arrive at the other side but also guaranteed arrive in order.</blockquote><p>28. some programming assignments are worth trying</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e00a32b6882f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-chpt-2-e00a32b6882f">Computer Networking — A Top-Down Approach Chpt 2</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Computer Networking — A Top Down Approach]]></title>
            <link>https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-ae1918768341?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/ae1918768341</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Sun, 19 Oct 2014 15:49:10 GMT</pubDate>
            <atom:updated>2014-10-23T13:58:13.478Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3P0x_6hKcYQW5U7sKSP5XA.jpeg" /></figure><h3>Computer Networking — A Top-Down Approach Chpt 1</h3><blockquote><em>CHAPTER 1</em></blockquote><p>Overall introduction to “Computer” network and Internet.</p><p>1. Network Edge — wire, wireless; DSL, Cable, FTTH, …; twisted-pair copper wire, Coaxial Cable, Fiber Optics, …;</p><p>2. Network Node — Packet Switching(queue, forwarding) <em>vs</em> Circuit Switching(frequency-division multiplexing (FDM) or time-division multiplexing (TDM))</p><p>3. Network of Network — tier-1 ISPs, lower-tier ISPs</p><p>4. Packet Switching — Delay, Loss, and Throughput</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/829/1*W3RrpKYqc5uoBmg1z0xXMA.png" /></figure><p>5. Nodal delay = nodal processing delay + queuing delay + transmission delay + propagation delay</p><p>6. Comparing Transmission and Propagation Delay:</p><blockquote><strong>The transmission delay </strong>is the amount of time required for the router to push out the packet</blockquote><blockquote><strong>The propagation delay </strong>is the time it takes a bit to propagate from one router to the next</blockquote><p>7. Queuing Delay —<em> La/R; L </em>bit-long packet; <em>R </em>rate transmission; a packet per sec</p><p>8. Packet Loss</p><p>9. End-to-End Delay — d = N * (d_proc + d_trans + d_prop)</p><p>10. Traceroute — Because of Queuing delay; round-trips_delay[i+1] &lt; rtd[i] is possible</p><p>11. Throughput — instantaneous throughput</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/996/1*QglYohArx7hBl5bdAXfoCQ.png" /></figure><p>12. Protocol layers; Layered architecture</p><p>a. Each layer performs certain action</p><p>b. Each layer uses service provided by the layer below it directly</p><p>13. Application Layer — Protocols: HTTP, SMTP</p><p>14. Transport Layer — TCP; UDP</p><p>15. Network Layer — Moving packets; datagrams</p><p>16. Link Layer — Ethernet; packet -&gt; frames</p><p>17. Physical Layer — actual bit transmission; medium of the link/bits</p><p>18. OSI — Application Layer, Presentation Layer, Session Layer, Network Layer, Data Link Layer, Physical Layer</p><p>19. Presentation Layer — Data compression, data encrytion, data description</p><p>20. Session Layer — delimiting &amp; synchronization of data exchange</p><p>21. Router VS Switches — $1 is able for IP protocol (3 layers); $2 only for 2 layers</p><p>22. DoS — Denial of Service</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ae1918768341" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/computer-networking-a-top-down-approach-ae1918768341">Computer Networking — A Top Down Approach</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Programming Pearls Clmn 14]]></title>
            <link>https://medium.com/programmers-dont-read-books/programming-pearls-clmn-14-70249e05ea8b?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/70249e05ea8b</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Mon, 04 Aug 2014 08:17:10 GMT</pubDate>
            <atom:updated>2014-08-04T08:40:00.139Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/409/1*m5-kArIM2J8VsUUPwZ3zZg.png" /></figure><p>Can you write a heap without a single error?</p><p>The author said we need 2 critical functions. One is siftUp, one is siftDown.</p><p>When I insert one element, I put it at the end of heap. Then shift it up until it’s proper location.</p><p>When I delete the top element, I put the last one of the heap to replace the first one. Then shift the new top down, until his proper location.</p><p>When we do heap sort. Do we need both siftUp and siftDown? No.</p><p>What we need to do is to build up the heap and delete the top element one by one, then maintain. And the building-up steps can be down using the maintain function.</p><p><a href="https://github.com/NTU-FSM/ACMaimeng-cheat-sheet#%E5%A0%86-heap">https://github.com/NTU-FSM/ACMaimeng-cheat-sheet#%E5%A0%86-heap</a></p><p>Here is a source code for heap sort.</p><p>Here is a important note you should remember:</p><blockquote>4. Heaps replace an O(n) process by a O(log n) process in all the programs.</blockquote><p>That is the ultimate charm of Heap.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=70249e05ea8b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/programming-pearls-clmn-14-70249e05ea8b">Programming Pearls Clmn 14</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Programming Pearls Clmn 13]]></title>
            <link>https://medium.com/programmers-dont-read-books/programming-pearls-13-8989e95b0113?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/8989e95b0113</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Thu, 03 Jul 2014 12:50:22 GMT</pubDate>
            <atom:updated>2014-07-03T12:55:51.772Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*L42ZH9-f0lOtf10wWgiOxQ.jpeg" /></figure><h4>Search where to write a blog? Haha!</h4><p>How many ways to do a search?</p><ol><li>use library</li><li>linear structure: linear search and binary search</li><li>binary search tree (BST): SBT, Splay Tree, AVL Tree, Treap and RBT</li><li>Bins</li></ol><p>We have see a lot about linear and binary search using linear structure. We should pay attention to those we are not familiar with.</p><blockquote>Tree</blockquote><p>I suggest that one should know all these tree structures and can write at least 2 of them immediately. That’s the way of study tree structures.</p><p>Maybe write some ACM problems in the searching category. Use multiple solutions to solve one classic question. ^_^</p><blockquote>Bin</blockquote><p>I want to admit that <em>bin</em> is a tricky structure. I made a easy question for my OIer mate. It is a structure-oriented question. There will be a lot of duplicated number in a list. All of the numbers are not greater than 100000, maybe. And you need to solve the list first and then make some decision. The list is not available to be sorted using quick-sort within the limited time. So, our best choice is <em>bin</em>, or you can also use a simple array — one number keep one slot.</p><p>One bad thing is that, I could now see any wide use of <em>bin</em>s in other application situations. It could because that I know too little. Anyway, still, <em>bin</em> is a good choice when you want to do searching.</p><blockquote>Library</blockquote><p>Hmm, library is good. Library is necessary.</p><p>But I insist not using these powerful libraries during learning. You use TreeMap, you use HashMap, but you know neither.</p><p>I suggestion learner should go through C programming language and then go to read <strong><em>Pointer on C</em></strong> and then learn algorithm and data structure using C or C++. That’s the way of learning. Definitely not using Java — Learn Java syntax, learn how to read Java doc. Then you are powerful enough by using thousands of libraries and an IDE…</p><p>Another way is that you know <em>Lisp</em>, <em>Scheme</em> or <em>Haskell</em> and then go to read <strong><em>Structure and Interpretation of Computer Programs</em></strong>. I admit that I do not know this book much. But I will read it may be during next semester. Not too far later.</p><p>I write this review after quite a long time after I finished this book. Because I went back home……T_T</p><p>I have to say: Home, Sweet home, You made me do not want to study at all!!!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8989e95b0113" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/programming-pearls-13-8989e95b0113">Programming Pearls Clmn 13</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Programming Pearls Clmn 12]]></title>
            <link>https://medium.com/programmers-dont-read-books/programming-pearls-clmn-12-3569f4d7323?source=rss----d2f7441a9429---4</link>
            <guid isPermaLink="false">https://medium.com/p/3569f4d7323</guid>
            <dc:creator><![CDATA[Lu Shengliang]]></dc:creator>
            <pubDate>Tue, 17 Jun 2014 09:22:45 GMT</pubDate>
            <atom:updated>2014-06-17T09:22:41.626Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/580/1*ZQBq19hUezFei08Bx7rAQg.jpeg" /></figure><h4>A “Simple” Sample Problem</h4><p>Okay, So what is the <strong><em>Simple</em></strong> Sample problem?</p><p>We come out with trying to solve a drawing a random sample from a list.</p><p>Our core problem becomes how to generate m distinct numbers within n.</p><p>Three solution are given for this problem. But you see, there is no prove for any of these algorithm. In stead, what we see here is just time complexity and space complexity, in additional, the code complexity.</p><p>Here we can figure out that the author is a completist. He required his student to be come as simple, as faster as possible.</p><p>This column is much easier to read than TAOCP(<em>The Art of Computer Programming, </em>Donald Knuth). I tried to read this book when I was year 1…</p><p>That book is a real killer. I think I should try this book instead of that one. This column is a good hook for TAOCP Volume 2.</p><p>(But that book is really heavy and expensive, LoL)</p><p>The problems are not to difficult, but some thoughts are worth borrowing.</p><blockquote>1. define new random int generation function</blockquote><p>Use what we get just now to solve what we will face in the future.</p><blockquote>6. The more you submit, the less mark you will get</blockquote><p>Honestly speaking, I like geek lecturer. Your lessons will be really funny. But I really do not like him/her to be the one who gives mark to your course work. Sometime you may get a slip under his/her geek instruction. Or he/her may think your work is not clever enough…</p><blockquote>7. Functional Programming</blockquote><p>Code in <em>Programming Pearls</em> are coded in pseudo code or C. TAOCP uses MIX. I still insist we need to read a algorithm book written in functional programming.</p><p>Maybe this is a good open source project:</p><p>Haskell and Algorithm. LoL</p><blockquote>8. Solution gives an excellent code</blockquote><p>The code is subtle. One could need read it several times to get the programmer’s idea.</p><p>One important thing here.</p><p>Code complexity, Time complexity and Space complexity are in a sophisticated trade-off relation. Which one is the most important one?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3569f4d7323" width="1" height="1" alt=""><hr><p><a href="https://medium.com/programmers-dont-read-books/programming-pearls-clmn-12-3569f4d7323">Programming Pearls Clmn 12</a> was originally published in <a href="https://medium.com/programmers-dont-read-books">Programmers Don’t Read Books</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>