id
stringlengths 7
12
| sentence1
stringlengths 5
1.44k
| sentence2
stringlengths 6
2.06k
| label
stringclasses 4
values | domain
stringclasses 5
values |
|---|---|---|---|---|
train_127100 | Move operations can be used in the context of modi.cations of the .ow structure, which occur very often. | exhaustive checking of alternative di.erences would lead to intolerable overheads. | reasoning | Software and its engineering |
train_127101 | The .rst page of an object stores not just data but also the frequently modi.ed object header which is why we avoid sharing this page. | our system passes R objects smaller than two pages of memory to the standard, non sharing memory allocator. | reasoning | Software and its engineering |
train_127102 | In the lookup table, the gain of .nding the physical address of replacement block is amortized by the number of OOB data reads incurred during a backwards search to locate a required page. | the advantages of the lookup table are smaller in a .ash with larger block size, since a larger block holds more replacement pages compared to a same size .ash with smaller block size. | reasoning | Software and its engineering |
train_127103 | In addition, a proxy object might be accessed via re.ection, e.g., by reading the .elds of another object. | the analysis treats all objects obtained that way as possible proxies. | reasoning | Software and its engineering |
train_127104 | In terms of data structure support, SAP Gravity is more .exible than Apache Wave or ShareJS since it allows syncing graph models that may include cycles. | it is capable to accommodate arbitrary business processes that can be expressed in BPMN. | reasoning | Software and its engineering |
train_127105 | For the sixth occurrence, we need to .nd CHR constraints of the form leq( ,X) to ensure that B = B ' holds. | we need an index on the second argument. | reasoning | Software and its engineering |
train_127106 | This contrasts sharply with the physical sciences, in which, for example, a measurement of the electron s charge would have an explicit statistical confidence bound arising from the intrinsic sensitivity of the equipment used in generating the evidence. | it is often difficult to use statistics to combine measurement error and credibility into a confidence bound around one estimate of a software attribute. | reasoning | Software and its engineering |
train_127107 | An additional issue with the hardware-assisted approach is that the necessary hardware mechanisms are typically speci.c to a particular micro-architecture. | exploiting them in a virtual machine that must ship on several architectures (and an even larger set of micro-architectures) represents a signi.cant engineering investment. | reasoning | Software and its engineering |
train_127108 | By Lemma 2, CSowner has the value p in some configuration C that lies between the steps Line(p, k, 1) and Line(p, k, 7[last]). | if the value of CSowner is not p in the configuration D, then there must exist a step s' in the fragment of the run from C to D such that s ~ changes the value of CSowner from p to something else. | reasoning | Software and its engineering |
train_127109 | In order to analyze the result in a better way, it is relevant know the pro.le of the assessors. | some characteristics as, age, gender, education level, knowledge in technology, expert or naive in video quality test, etc. | reasoning | Software and its engineering |
train_127110 | This is mainly because of the memory accesses and fewer numbers of blocks needed to fully utilize the GPU. | tuning stops increasing the aggressiveness of the optimization because it does not provide any further speedup. | reasoning | Software and its engineering |
train_127111 | Second,ineach classversionthestaticinitializerisextended toinvoke the pbi_initClass() methods of all class versions inthe end(if thereis no staticinitializerin a class version,itis created). | after merging of the staticinitializers, the pbi_initClass() methods of all class versions willbe executed, independently ofthe executed code version ofthe merged staticinitializer.Thatis,allinsertedstatic .eldswillbeproperlyinitialized. | reasoning | Software and its engineering |
train_127112 | Hence, we simply add the following line to the tracematch de.nition, giving the clue that actions occur more frequently than other symbols: frequent action ; Association aspects also choose their indexing structure based on the look-up direction. | look-up is guaranteed to be fast. | reasoning | Software and its engineering |
train_127113 | The class E.F could then be an extension of Top.this.a.b.C. | only if there is a common class wrapping the classes in point and there is a path of .nal .elds to the class that is to be extended. | reasoning | Software and its engineering |
train_127114 | MEASUREMENT AND RESEARCH ENVIRONMENT Tiwari mentioned in his paper that time the processor used was directly related to the power it needed [3]. | to get the same output from similar essential working steps while controlling other kinds of element, the shorter the processor time uses the better performance of the chosen component is - in term of the power optimization. | reasoning | Software and its engineering |
train_127115 | Abductive exploration just performs the function of a model builder that should be followed for a con.rmatory data analysis. | abduction plays the role of an explorer of viable paths, but further inquiry should determine which are the most plausible solutions [43]. | reasoning | Software and its engineering |
train_127116 | But, the principal contribution of Little-JIL is perhaps its suggestion of abstractions that seem particularly effective in communicating process thought. | little-JIL is a member of a newer family of process definition languages aimed at determining what these models and abstractions need to look like. | reasoning | Software and its engineering |
train_127117 | In other words, the dynamics of any given software development effort make it necessary to revise written requirements. | waterfall approaches are faulted for not allowing for revision. | reasoning | Software and its engineering |
train_127118 | This constraint means that, viewed as unsigned 32-bit integers, I1 is strictly less than I2. | we view the machine as capable of manipulating mathematical integers,but the comparisons are limited to a 32-bit perspective on these mathematical integers. | reasoning | Software and its engineering |
train_127119 | Toautomatethe entireloadingprocessitshould runwithout user interaction. | for dynamic loading of appropriate bundle code at runtime, bundles have to be describable with metadata characterising the OSGi bundle/service requirements and dependencies (see Figure 2). | reasoning | Software and its engineering |
train_127120 | In a work-stealing program, tasks often have the same features (e.g., number of instruction, compute-intensive or data-intensive). | the workers of a work-stealing program have similar demands on the computation resources as well. | reasoning | Software and its engineering |
train_127121 | An instruction needs to be dispatched before it can execute, and execute before being committed. | within each instruction s nodes there is a DE edge and an EC edge. | reasoning | Software and its engineering |
train_127122 | Furthermore, modern supercomputers are often configured by connecting multiple GPUs in large-scale networks due to the advantages of such architectures, including their very high cost/performance ratios and low power consumption. | outside of the GMs, it is also essential to have some way of detecting faults in the interconnections between the SMs and GMs, or in the network between the GPUs. | reasoning | Software and its engineering |
train_127123 | ECMAScript 5 s frozen objects help to simulate ActionScript 3 classes more accurately. | jangaroo will not become dispensable, but on the contrary, it will be possible to approach full support for Action-Script 3 in modern browsers. | reasoning | Software and its engineering |
train_127124 | Disabling network bu.ering for puts (this happens to be correct for this benchmark) brings the auto version s performance to within 3% of that of manual blocking, but has no effects on the other benchmarks. | while our framework is e.ective for small to medium transfer sizes, a zero-copy consistency checking algorithm could further bene.t applications with large-sized puts. | reasoning | Software and its engineering |
train_127125 | Only when they are equal, the proof passed in by the user can be casted to a proof of pf (Cashier says (Paid m)), where m is the value of the stored current nonce. | an old proof with an expired nonce will not reveal the secret. | reasoning | Software and its engineering |
train_127126 | As you can see in Figure 10, the store replication itself worsens the performance of the program. | we need to cancel such a transformation if the loop cannot be transformed to the faster code. | reasoning | Software and its engineering |
train_127127 | Although thebenchmarks, when partitioned into tasks, do not contain inter-task dependences, false sharing introduces dependences that occasionally cause speculation to fail. | webelieve that ourresults re.ect theperformance ofbenchmarks with dependences. | reasoning | Software and its engineering |
train_127128 | In contrast to R.1, we also need to modify annotations on members, because we include the containing class in more variants than before. | we propagate the classes annotations to their members.3 Similar to merging classes from mutually exclusive features, with refactoring R.3, we also merge members with the same name inside a class. | reasoning | Software and its engineering |
train_127129 | If we were to lock the data structure for each of Q queries, each query might be forced to wait for insertions and other queries, thereby increasing the apparent work by as much as T(QP) and nullifying the advantages of P-way parallelism. | we lock the entire global tier when an insertion occurs, but use a lock-free implementation for the presumably more-numerous queries. | reasoning | Software and its engineering |
train_127130 | We felt these issues would make it dif.cult to extend Ruby s parser for our own purposes, e.g., to add a type annotation language for DRuby. | we opted to write a Ruby parser from scratch. | reasoning | Software and its engineering |
train_127131 | For example, collocating two blocks in a 13region, which includes two virtual pages in our system, produces a 50% probability that two blocks are in the same page. | other values of K affect the probability for cache or page collocation. | reasoning | Software and its engineering |
train_127132 | Consequently, different tools are able to accurately render and edit the globally traced artifacts. | a new artifact can be entered via a Word document, modified through a web data entry form, and viewed in a table-format report. | reasoning | Software and its engineering |
train_127133 | If we consider a product line with hundreds and thousands of features [1], it becomes obvious that product-based measurement does not scale. | we need better methods for determining quality attributes of all products. | reasoning | Software and its engineering |
train_127134 | 3) Finally, smartphone users can benefit from a more flexible choice of sensors for their intended applications, as long as the sensors have the Dandelion support. | users will also benefit from a wider range of emerging applications and sensors. | reasoning | Software and its engineering |
train_127135 | Second, if a variable is allowed to have multiple lifetimes, they may be assigned di.erent colors, depending on the order of the basic blocks. | in connecting the data.ow, the number of data.ow transfer instructions needed is also dependent on the order. | reasoning | Software and its engineering |
train_127136 | This implies that this value can be recovered neither by the rede.ne technique nor by the extract-from use technique. | the RCG algorithm resorts to state saving which is performed by the inserted save instruction before i1. | reasoning | Software and its engineering |
train_127137 | At a call site to the procedure, the objects accessed via the iterator are resolved to be parts of composite objects. | applying the summary of the procedure results in some points-to relations which have the pointer components of composite objects as sources. | reasoning | Software and its engineering |
train_127138 | Note that if two or more input (output) transitions have the same output (input) vector, we can fuse them in one transition whose rate is the sum of the rates of the original transitions. | without loss of generality, we assume that all the input (output) transitions have di.erent output (input) vectors. | reasoning | Software and its engineering |
train_127139 | We assume in this work that the collectors operate on spare processors. | complicated scheduling issues that often appear with uniprocessors do not arise. | reasoning | Software and its engineering |
train_127140 | 2) An atomicity violation may be benign it does not affect the correctness of the program. | in order to understand the effectiveness of the tool, we tried to identify if an atomicity violation error reported by ATOM-FUZZER could lead to a bug. | reasoning | Software and its engineering |
train_127141 | However, as mentioned in Section 5.2, whenever a formula is satis.able there always exists an in.nity of possible labellings which satisfy it. | rather than proposing just one labelled tree, the solver gives a minimal tree together with labelling constraints representing all labellings which make that particular tree a counter-example. | reasoning | Software and its engineering |
train_127142 | Additionally, Figure 5(d) con.rms that the latter strategy promises a better temporal workload balance. | thelatter strategyispreferred to cope withpeerheterogeneity. | reasoning | Software and its engineering |
train_127143 | Since routing nodes in a wide-area pub-sub system are typically managed by entities in di.erent administrative domains, we must assume that publishers do not trust routing nodes in other domains in terms of the con.dentiality of their published data. | our secure aggregation protocol should allow untrusted routing nodes to perform aggregation of raw data without learning the private input values. | reasoning | Software and its engineering |
train_127144 | In other words, xi = cji for 1 =i =k. | since each count c is independently selected with probability g(c), for each occurrence of c, on the average g(1 c) -1 other nodes have (I,c)but did not sample it; that is, each occurrence of ccan be thought of as representing g(1 c) occurrences of c. we estimate the total count fI of I as x1 x2 xk fbI = + + ···+ (1) g(x1) g(x2) g(xk) The following theorem shows that this estimator is unbiased and establishes its MSE. | reasoning | Software and its engineering |
train_127145 | In order to continue exemplifying the use of negative scenarios let us assume that the new implied scenario is also deemed as undesired behaviour by stakeholders. | we obtain a new negative scenario (Figure 10) from which a new constraint can be built to check for further implied scenarios. | reasoning | Software and its engineering |
train_127146 | First, for each considered parameter, the values of that parameter in the two samples are being compared; a parameter similarity factor (p SimF) is calculated for each such parameter. | there will be an individual memory usage p SimF, a workload p SimF and so on, for each monitored parameter. | reasoning | Software and its engineering |
train_127147 | However, a merge combines a data block and a log block into a new data block. | physically, a merge results in using an already erased block but producing two new erased blocks. | reasoning | Software and its engineering |
train_127148 | Barring special thread interfaces such as OpenMP, however, the number of threads cannot be dynamically adjusted during runtime for most workloads. | if the power cap or workload behavior vary signi.cantly during runtime, it is highly likely that the system operating point will not lie on the power-delay Pareto ef.cient Table 2: A comparison of the average power and execution delay of our MLR classi.er with and without thread packing for a 130W power cap. | reasoning | Software and its engineering |
train_127149 | On the other hand, a better bound has been proved for C B C -M AC than for our construction, namely on the order of 2-blq2 [7] (when l < 2b/3). | the trade-off between slightly slower performance and better bound should be explored. | reasoning | Software and its engineering |
train_127150 | Then the derivation ends with TEXP followed by an application of subsumption for the re.exive case. | f f e1 : L and f f e2 : L, and by induction µ(e1)= v(e1) and µ(e2)= v(e2). | reasoning | Software and its engineering |
train_127151 | Such a script can direct other applications to perform useful functions, but may also call other script .les. | script .les provide a simple mechanism for modularity. | reasoning | Software and its engineering |
train_127152 | In addition, we also remove the user provided atomicity constraint of incAndLog,sonow M3 (mutator increments) and mutator logging are no longer constrained to execute atomically. | the only atomicity constraints left in the mutator are those forming the individual building blocks. | reasoning | Software and its engineering |
train_127153 | However, this comes at a higher storage cost in memory, as well as higher mem- ory traffic. | the rest of this subsection gets into greater detail comparing DP with RP, in terms of per- formance implications of these prediction accuracies, par- ticularly for the applications with higher TLB miss rates. | reasoning | Software and its engineering |
train_127154 | If we do succeed, the new coloring can be committed. | we overwrite the old color with the temporary one and reset the changed .ag for each node in the list. | reasoning | Software and its engineering |
train_127155 | We deliberately chose to measure the gross effect including all startup and transfer overheads as that is the impact that eventually can be observed by programmers in terms of wall-clock runtime effects. | we present speedups of generated CUDA programs (referred to as SAC-CUDA) against their sequential counterparts (referred to as SAC-SEQ). | reasoning | Software and its engineering |
train_127156 | Working independently, these algorithms suit different types of programs. | using one or more of these algorithms, a variety of programs can bene.t from the use of Of.ine GC. | reasoning | Software and its engineering |
train_127157 | Sometimes design patterns introduce complexity. | before applying the transformation process, the impact of pattern oriented refactoring on quality attributes[11] needs to be estimated. | reasoning | Software and its engineering |
train_127158 | Note that Permissions propagated upwards via a doPriviledged node do not propagate beyond the predecessor of the doPrivileged node. | the above definition (or the algorithm based on it) must be refined to replace G-() by G- n, p), where n ( To clarify all of the concepts introduced in this section, consider the following simple example. | reasoning | Software and its engineering |
train_127159 | To achieve maximum degree of semantic interoperability, the SNOMED CT ontology is one of the choices to use in standard MLMs. | the use of standard information model HL7 vMR and standard terminologies such as SNOMED CT ontology in MLM provide interoperability and exchangeability features in knowledge base. | reasoning | Software and its engineering |
train_127160 | However, in terms of emulating the NUMA API, while these results show slightly more variation between the two implementations, the differences, again, are always within the margin of error. | our framework provides an effective implementation of the NUMA optimization in HotSpot. | reasoning | Software and its engineering |
train_127161 | High-level languages provide type-safety, memory-safety, encapsulation, and strong abstraction over hardware [12], which are desir able goals for system programming as well. | high-level languages are potential candidates for system programming. | reasoning | Software and its engineering |
train_127162 | At room temperature for silicon, fv is 15.6 THz and the ratio of power for stoke and anti-stoke output is around 10. | in this paper, we will only consider the stoke wave for maximum power emission from the Silicon Raman laser. | reasoning | Software and its engineering |
train_127163 | In languages such as Java in which objects are always accessed indirectly through references, points-to information is also an important prerequisite for many analyses, and points-to analysis is often interdependent with call graph construction. | tools for comparing points-to analysis results should also be developed, perhaps combined with the call graph comparison tools. | reasoning | Software and its engineering |
train_127164 | (R ) The secondhypothesis of(R ) creates O(#edge.2) table entries, in contrast to O(1) table entries created by (L ). | we conclude that the right-recursive version should run a constant factor slower.The experimentsinSection6 con.rm this. | reasoning | Software and its engineering |
train_127165 | In other words, the value of the corresponding logical expression must be changeable through some user interface control. | our approach works in two steps: 1) Identify all the alternative paths to the backbone execution. | reasoning | Software and its engineering |
train_127166 | In the next O(log k) time units the win- ner of the tournament tree is determined (according to the properties of the tournament tree). | in O(log k) time units some process enters the critical section. | reasoning | Software and its engineering |
train_127167 | Change in data correlations may result in changes to the correlation graph. | each deleted node s periodically checks for validity of the hyperedge (S, s)usedin its C3 condition by gathering data from the nodes in S (using the piggyback strategy over normal data-gathering process as described in Section 2.3), recomputing the a parameters (Equation (2)), and checking if the least square error (Equation (1)) is within the given bound. | reasoning | Software and its engineering |
train_127168 | Generally, private cloud platforms may benefit from a better network interconnect, better security, and more user control over the resources. | private cloud platforms are ideal for testing scientific applications and identifying best practices for scientific computing on cloud platforms. | reasoning | Software and its engineering |
train_127169 | Therefore, the difference between nx (where every ai =1) and R increases, and the given range increases. | the effect of using virtual nodes will increase as the system becomes more heterogeneous. | reasoning | Software and its engineering |
train_127170 | The former is comparable to the Object... parameter list in Java: variadic arguments can be accessed through void pointers, but also their typeid is available. | the programmer can safely cast the arguments to their expected types and deal with errors in a controlled manner. | reasoning | Software and its engineering |
train_127171 | However, it still requires all non-deterministic events, even those related to messages exchanged inside a cluster, to be logged during failure-free execution to ensure that the exact same execution can be reproduced after a failure [7]. | the performance of these protocols can also be impaired by the overhead of event logging. | reasoning | Software and its engineering |
train_127172 | 0 if |f1 - f2|= a I(f1,f2,a)=(|f1 - f2|) (1) 1 - otherwise a For IEEE 802.11b/g three channels of the 2.4GHz frequency band are theoretically orthogonal. | we set a to 30MHz, which results in the following set of orthogonal channels {1, 7, 13}. | reasoning | Software and its engineering |
train_127173 | In this paper, we are mainly interested in how this choice of base language affects the design of aspect-oriented features. | we present an aspect-oriented extension to Petri nets and demonstrate its use with an invariant enforcement example. | reasoning | Software and its engineering |
train_127174 | Thisleaves alotofBluetoothbandwidth unused because the Bluetooth interface in BlueStreaming is always active. | when there are uploading packets pending, un-used Bluetooth bandwidth is .rstly exploited to transmit the uploading traf.c. | reasoning | Software and its engineering |
train_127175 | 3.3 Fault Localization An EFG is a Bayesian network which models the error propagation in a set of program runs. | we can apply an inference technique on the EFG to .nd the root cause of the error. | reasoning | Software and its engineering |
train_127176 | This highlights that a single ISTOF value (maximum stress traffic) can happen in different time instances, thus suggesting the search landscape for the GA is rather complex for this type of problem. | a testing strategy to further explore would be to cover all (or a subset of) such test requirements with maximum ISTOF values in different time instances. | reasoning | Software and its engineering |
train_127177 | THE FUNCTIONAL PARADIGM The definition of an MPL-RM function is required to be free of side effects. | neither global variables nor objects may be destructively updated within a function. | reasoning | Software and its engineering |
train_127178 | We have implemented this idea by creating a predefined concrete interface for each kind of traditional constructor, which extends the set of operations provided by the constructor. | the record constructor is associated with a predefined concrete interface with well-defined semantics. | reasoning | Software and its engineering |
train_127179 | Taking a snapshot when an object has been converted violates the third safety requirement for using a snapshot from an optimized execution in a debug build. | we restrict the compiler to only convert objects that are non-escaping and whose lifetimes do not cross snapshot safe points. | reasoning | Software and its engineering |
train_127180 | However, the relevant sets of objects may not be explicitly available in the original program. | in this paper, we present a declarative approach for de.ning sets of objects with the same usage pattern by means of a dedicated language construct. | reasoning | Software and its engineering |
train_127181 | All graphs of order lower than 11 are also included in this amount. | any possible graph up to Order 11 can be pre-calculated and all instances of known character graphs can be mapped to these pre-calculated models. | reasoning | Software and its engineering |
train_127182 | This is a hard problem both in theory and in practice [2]. | rather than compute the pre cise answer, we seek to bound the probability of satisfaction. | reasoning | Software and its engineering |
train_127183 | In our solution, we exploit the interpretation that ensembles can be viewed as a con.guration of a .attened SPL reference architecture. | we create an SPL con.guration by (i) transforming the SPL reference architecture to an architecture of DEECo components and ensemble de.nitions, and (ii) re.ecting the establishing and dissolution of DEECo ensembles at the level of the SPL reference architecture. | reasoning | Software and its engineering |
train_127184 | At the end of each round, the results in array tempU are copied back to U.The (ij)T loop nest implements the actual 5-point stencil computation, and is the focus of our approach. | we will omit the outermost t loop and the array copying part (which is not a stencil computation) in the rest of our discussion. | reasoning | Software and its engineering |
train_127185 | Because of these problems, it is di.cult to use an only PRAM main memory. | researchers consider a hybrid main memory architecture with DRAM and PRAM. | reasoning | Software and its engineering |
train_127186 | As a result, based on the concavity of R(ek)inTheorem2and the optimalityof Theorem1,wecandividetheoverall optimization problem into two subproblems, says inter-frame and intra-frame energy assignments. | what we have to do is to determine the intra-frame energy assignment.Byapplying the Lagrange Multiplier method, we can solvethe problem in Equation (1) optimally, as shown in Algorithm 2. | reasoning | Software and its engineering |
train_127187 | However, with limited profiling, sophisticated path profiling schemes can no longer offer any prediction advantages over simpler schemes that exhibit much lower runtime overheads. | while intuition may call for longer and more elaborate profiling, we will show in fact show that the opposite it true: less profiling actually leads to more effective predictions. | reasoning | Software and its engineering |
train_127188 | In fact, the least .xed point assigns to each port the empty signal. | deadlock manifests itself as a least .xed point of empty or .nite sequences. | reasoning | Software and its engineering |
train_127189 | We assume that the security manager establishes an aggregation path such that there between pi and psub. | at least m + 1 colluding parties are required to obtain the value of vi. | reasoning | Software and its engineering |
train_127190 | By the previous rollback scheme, the federate state is always recovered at the point of an NMR call to the TiMaC layer. | this means that, upon state recovery, the federate automatically re-starts from the epoch with initial time trecovery .after having performed the state recovery phase through the Recovery-Action B, TiMaC skips execution of Step 1 and immediately goes on executing Step 2 and Step 3 for that epoch. | reasoning | Software and its engineering |
train_127191 | i=1 where n is the number of .elds of o; Lo is the lifetime of th instance o; li is the lifetime of the i.eld of o; r(o)is the idle .eld ratio for o; b is the size of each .eld; and Bo is the total size of o (i.e., Bo = nb). | p(o) is determined by three factors: the idle .eld ratio, size and lifetime of instance o. | reasoning | Software and its engineering |
train_127192 | While both approaches are potentially good, the .rst one is likely to be harder and, in addition, not very e.ective when the algorithm runs with a small number of processors like in our experiments. | in this paper, we focus on the second approach. | reasoning | Software and its engineering |
train_127193 | automaticdeductionofacompoundexpression ssemanticvalue:Ifthecompoundexpressionreferencesonlyasinglenon-voidnonterminal,thatnonterminal ssemanticvaluemustbetheoverallexpression svalue. | atext-onlyproductionisaproductionwithadeclaredtypeofString.Additionally,itmustnotcontainassignmentstoyyValue andmayreferenceonlyothertext-onlyproductions.Thesemanticvalueofatext-onlyproductionisthetextmatchedintheinput.text-onlyproductionseliminatetheneedforexplicitlybuildingupstringsthroughJava sstringbuffers;onasuccessfulmatch,theimplementationsimplycreatesthestringfromthebufferedinputcharacters.Text-onlyproductionsaretypicallyusedforrecognizingidenti.ers,keywords,andliterals.Forexample,thefollowingtext-onlyproductionfromourJavagrammarrecognizesstringliterals: String StringLiteral = ["] (EscapeSequence / ! | reasoning | Software and its engineering |
train_127194 | Furthermore, clients also have to pay money for transferring data across the networks to/from the providers. | depending upon how much the client is willing to pay, the client might want to do some processing at the server in order to reduce the amount of data that it has to transfer to the client. | reasoning | Software and its engineering |
train_127195 | Traditionally, clean-slate design tends to focus on creating effective and elegant networks for a static set of requirements. | the approach is often one of engineering, which tends to neglect competitive effects. | reasoning | Software and its engineering |
train_127196 | Additionally we assume that time constraints allow us to execute only n% of the original test suite. | lru(n) chooses the n% of the test suite with the highest selection probabilities. | reasoning | Software and its engineering |
train_127197 | Moreover, the other application processes will be able to use their local checkpoint copies. | this storage mode provides recovery as long as one of the two nodes containing a checkpoint replica is available. | reasoning | Software and its engineering |
train_127198 | Given variables tl, ..., tl+n .fv(C), representing the types of locations l, ..., l+ n,if a.fv(.ti), then the type at location i contains the variable a. | we report it as a candidate for a type annotation. | reasoning | Software and its engineering |
train_127199 | DEFINITION1(CAPSULESTATEHART).Aapsule statehartSC=(S,E,A,T,P,port)onsistsof anon-empty,fnitesetSofstates, afnitesetEofevents, afnitesetAofations, atransitionrelationTSxExAxS,where e/a(si,e,a,s})ETisdenotedbysi -s},and afnitesetPofportnames. | wedistinguishbetweeninternalandexternaleventsandations.Externaleventsandationsareausedbyreeiving andsending,respetively,amessageviaaport.we assumeafuntionport:E+A-P+{I}assoiatingto eaheventorationxaportnameinPorthedistinguished labelIdesignatinganinternaleventoration. | reasoning | Software and its engineering |