id
stringlengths
7
12
sentence1
stringlengths
5
1.44k
sentence2
stringlengths
6
2.06k
label
stringclasses
4 values
domain
stringclasses
5 values
train_127000
In this regard, in order to reduce the number of tests cases and save time, we found that tests also should be oriented to features.
this would facilitate executions of test cases for both generative and runtime techniques.
reasoning
Software and its engineering
train_127001
One approach to improve the detection of redundant in­formation for non-WinZip encodings is to reduce the aver­age chunk size, but even using CDC with an average chunk size of 2 KByte, we have only been able to detect 51.3% of redundant information.
we have created an application-speci.c chunking method for zip-.les.
reasoning
Software and its engineering
train_127002
Blackscholes shows good scalability at least up to 32 threads and does not show vari­ation in the right number of threads to run even during the course of execution of the program.
lIMO picks 32 threads this benchmark does not exhibit varying characteristics during its execution.
reasoning
Software and its engineering
train_127003
This change also influences the execution of Class B since it inherits the changed Class A.
all the methods of Class B are also marked as changed.
reasoning
Software and its engineering
train_127004
Firm mutation is de.ned by Woodward and Halewood [24] as the situation where a simple error is introduced into a program and which persists for one or more executions, but not for the entire program execution .
similarly to the AjMutator tool [10], the evaluation of mutants based on partial program executions implemented in JUnit tests characterises .rm mutation in Proteum/AJ .
reasoning
Software and its engineering
train_127005
Note that concurrent UN requires about 28% more heap space than STW UN (Section 4.6).
bounding space overhead in concurrent MC does not seem neces­sary/practical.
reasoning
Software and its engineering
train_127006
(3) If communication c is performed right after communication (c - 1), this means that an output sent by a node follows multiple links before it reaches its destination.
the communication in the .rst link (c - 1) has to start before the next one (c).
reasoning
Software and its engineering
train_127007
Clearly, reading through each non-sort-key attribute of each tuple in a entire large result set is a slow and ardu­ous task.
our task becomes that of supporting fast browsing (and more speci.cally, scrolling) over such a sorted relational result set.
reasoning
Software and its engineering
train_127008
Note that, before the context transition, GPSNavi is still active and RFNavi is not active yet.
the execution of powerOffDevices() results in the execution where the device controller turns the GPS receiver o.. By EventCJ, we can declaratively specify which context is activated by which event, and this activation is preserved until the next event that deactivate the context is .red.
reasoning
Software and its engineering
train_127009
credential, declaration and policy files) have been designed for TN, the file format interoperability problem continues to be a challenge.
approaches for the establishment of a trust relationship are required to further support communication efficiencies between WS.
reasoning
Software and its engineering
train_127010
Moreover, in principle all approaches need to address concepts like resource demands, consumption costs, quality mappings and selections of configuration parameters, but they don t usually explicitly focus on or expose their descriptions and their relationships.
it would be interesting if one could easily separate the modeling of these concepts, the relations between them, and their usage, from the engineering and technology details selection level, in order to be feasible to easily construct realizations of different system-wide architectural alternatives.
reasoning
Software and its engineering
train_127011
The discussion concerns both the modeling language underlying the approach and the steps taken in the approach.
we prepare the ground for the road map.
reasoning
Software and its engineering
train_127012
Like Knuth, they argued that their tools could bene.t by knowing the typical and extreme values of various aspects of software.
their interest is in the low-level details of the code with a view toward future tool support or lan­guage design.
reasoning
Software and its engineering
train_127013
We don t care about J s knowledge.
the set of agents A = {S, P }.
reasoning
Software and its engineering
train_127014
We assume that the information bits have already been compressed, such that for any subsequence of length n> 0, the 2n possible sequences are sent by the sender with equal probability.
no set of information bits sent by the sender contains useful information about other bits.
reasoning
Software and its engineering
train_127015
4.3 Workloads creation To exhaustively validate our thesis, we tested all the possible workloads featuring a number of instances for both bodytrack and ferret ranging from 0 to 4, imposing that the total number of appli­cations would be greater than three.
there is always at least one binding domain where two applications share the resources.
reasoning
Software and its engineering
train_127016
In the present paper, we focus on examining the effectiveness of the studied strategies with respect to a set of mutants.
the use of the above restrictions affects only the initial set mutants and not their impact assessment.
reasoning
Software and its engineering
train_127017
One then requires that .nally clauses cannot discard errors (unless maybe, they throw an error themselves).
for example, if a try (or catch) block throws an error E, the .nally block is executed.
reasoning
Software and its engineering
train_127018
Starting with a large set of potentially relevant studies, reviewers rely on several steps of selection, .rst by reading titles and abstracts, followed by full-text reading for quality assessment [3].
the reliability of a secondary study is highly dependent on the repeatability of the selection process [8].
reasoning
Software and its engineering
train_127019
The interaction takes place only between the system and its environment and between the subject and the consid­ered system, respectively.
the syntactic interfaces of U and E are denoted by (OU I IU ) and (OE I IE ), respec­tively.
reasoning
Software and its engineering
train_127020
Once an object becomes thread-shared, a process which must be signi.ed by the share annotation, it can no longer be mentioned in another object s invariant.
examples like the one shown in Figure 13 where the invariant of the RequestPro­cessor class depends on the thread-shared RequestPipe ob­ject, cannot be veri.ed.
reasoning
Software and its engineering
train_127021
Many LOAD CONST in­struction occurrences load objects that have been allocated in the lower memory area below 32 bits.
for all constant objects where this observation holds, we can ac­tually replace the array indirection by a direct push of the instruction argument.
reasoning
Software and its engineering
train_127022
It should be emphasized that the pre-trusted peers are essential to this algorithm, as they guarantee convergence and break up ma­licious collectives.
the choice of pre-trusted peers is important.
reasoning
Software and its engineering
train_127023
T 2 -46 = 0 (eigenvalues equal and real): The variation of a+a u as a function of time can be calculated as u = u0e -t 2 , where u0 is the initial value of u.
the system converges exponentially quickly from the neighborhood of the second equilibrium point.
reasoning
Software and its engineering
train_127024
As discussed previously, if a bind handle can begarbage collected that signi.es that the advising relation­ship is never going to be removed.
it can be optimized away using techniques proposed by Bockisch et al.
reasoning
Software and its engineering
train_127025
Although we are investigating the discipline with a plethora of empirical studies, those stud­ies either concentrate on validating speci.c methods or on single companies or countries.
they allow only for limited empirical generalisations.
reasoning
Software and its engineering
train_127026
The difference in program performance between the feature-vector based heuristic and the focus-method GA is scaled with the fraction of overall program time spent in the relevant method.
we can see from Figure 8 that the per-method performance results achieved by the feature­vector based heuristic are quite disappointing.
reasoning
Software and its engineering
train_127027
For procedures with multiple return statements, it is inconvenient to add that code in multiple places.
a procedure is transformed such that it has only a single return statement.
reasoning
Software and its engineering
train_127028
3.3 System Implementation The first objective is to identify the possible problem.
one part of the system is to evaluate the user s state.
reasoning
Software and its engineering
train_127029
The feedback that usersprovide ononeversionofanapplication isautomaticallycar­ried forward, using defect differencing, to the next version.
the feedback-import feature can be used to customize future re­ports: Invalid and WontFix bugs keep recurring from one ver­sion tothe next, but can easily be .lteredout so that the user is not distracted by them.
reasoning
Software and its engineering
train_127030
Both are explicit metastates, which means that they contain user-defined action code.
the modeler is able to specify custom behavior for acceptance and rejection of submitted papers.
reasoning
Software and its engineering
train_127031
on selectors that are themselves unde.ned have the return type [a], which is again pointed without any condition on a.
launchbury and Paterson s approach to parametricity in the presence of pointed vs. unpointed types can be used to show that law (1) holds without any conditions on p, h, or l, provided all invocations of seq are dropped2 2By contrast, if, e.g., recursion were performed at a type whose (in which case we get precisely the .lter function from Haskell s standard prelude).
reasoning
Software and its engineering
train_127032
In an organization, for example, different stakeholders from different departments can perceive some opportunity for software depending on their knowledge of it and the added value they are looking for behind its adoption.
there is a need for tools to make business value more tangible and limit subjectivity issues.
reasoning
Software and its engineering
train_127033
Since the exit stub stores the CTI location, it is larger compared to the exit stubs used by proac­tive linking.
the auxiliary code size is larger for lazy linking than for proactive linking.
reasoning
Software and its engineering
train_127034
To eliminate half of the cross-core interference penalty of a is more valuable than b, meaning the bene.t gained, a 20% in­crease in speed, with a is better than the 1.9% speed up in b.
we should be willing to sacri.ce more utilization to eliminate 50% of the cross-core interference penalty for a than b since a is more cross-core interference sensitive.
reasoning
Software and its engineering
train_127035
This and Lemma 3.2 imply that the CurrSeq parameter of gv(S ' ) has the value written in Rs by S ' .Since gv(S ' ) reads the value written by U,it follows by the code that if U reads in Rs avalue for CurrSeq greater than or equal to the one written there by S ' , gv(S ' ) will terminate by executing line 26, which is a contradiction.
u reads in Rs avalue for CurrSeq less than the one written there by S ' .So, U starts its execution before S ' writes to Rs.
reasoning
Software and its engineering
train_127036
In other words, one loop iteration takes II cycles (ignoring the loop pro­logue and epilogue).
minimizing the II directly leads to an increased throughput.
reasoning
Software and its engineering
train_127037
However, if the com­ponent J modi.ed this data as shown in Figure 1c, then the read element would be authors and the obtained value is Single .
if the component J inserts a data element into the path to some elements accessed by the component C, then the result of interference of these operations is that the component C accesses and reads values of different data elements from what was intended when it uses sequence numbers of elements rather than their names.
reasoning
Software and its engineering
train_127038
As noted, such a generic aspect de.nes both a method (checkInv) and a pointcut (cePC) as abstract.
they a sup­posed to be de.ned by a concrete aspect which extends it.
reasoning
Software and its engineering
train_127039
C-DIDUCE isdifferentfromtheoriginalimplementationinthefollowingways:1.BecauseC isnotanobject-orientedlanguage,wetrackprogrampointswhichreadfromorwritetoglobalvariablesinsteadofobjects,andalsostatic-typedobjects;2.formulti­worduser-typedobjects(includingstruct),wecastthe.rstwordintoanintegervalueandusethisasthevalueoftheobject;3.sinceC hasnoobjectreferencesorinheritancerelations,weignorepointertypes;and4.bydefault,C-DIDUCE doesnotrelaxinvariantsduringthecheckingmode.Repeatedviolationreportspointingtothesamelocationaremergedintoone.Otherwise,C-DIDUCE isimplementedthesameastheoriginalDIDUCE (includingtheviolationrankingscheme).C-DIDUCE instrumentsthesourcecodewithcallstotheC-DIDUCE runtimelibrary(containingC-DIDUCE checkingandreportingfunctions).
tomeasuretheeffectivenessofdetectingvalueinvariantvi­olationswhenusingtheArtemisframeworkwithC-DIDUCE,wemeasuretheoverlapoftheinvariantviolationsintheSPEC 2000 benchmarksdetectedandrankedusingC-DIDUCE and{C-DIDUCE +Artemis}.Thisisjusti.edsinceC-DIDUCE (andDIDUCE)onlydetectandrankruntimeviolationsofinvariants theydonotac­tuallyidentifybugs.theextentthatArtemisimpactstherankingsgivenbyC-DIDUCE isindicatedbyhowmanyviolationsdetectedbyC-DIDUCE andrankedtopX isstilldetectedandrankedtopX whenusingC-DIDUCE withArtemis.
reasoning
Software and its engineering
train_127040
coin .ips, the number of distinct algorithms in LII is at most |O|n., where O is the domain of the coin .ips.
by Yao s Principle, for any distribution GII over LII , D is the event that b .nds a process in that phase.
reasoning
Software and its engineering
train_127041
In future work, we plan to investigate, model and eval­uate further aspects and their applicability to the domain of distributed service-oriented multimedia content analysis.
one of our targets is to automatically integrate cross­cutting concerns such as reliable messaging and pro.ling.
reasoning
Software and its engineering
train_127042
If there is a back­ground volatility in measured values, there is no indi­cation of statistical con.dence that the sample in fact indicates a systemic problem.
unneces­sary adaptations maybe performed leading to failures or increased costs associated with adaptation.
reasoning
Software and its engineering
train_127043
We use thisfactto start storingthedata itemsof function bar from the .rst unusedslotinthe chunk aslotthatwouldbeusedbyfunction foo but only after functionbar returns.
foo and bar share those four slots thatwould otherwise remain allocatedbut unutilized.
reasoning
Software and its engineering
train_127044
Reputations are often used to distinguish between users, and in many applications, users care more about relative reputation values than the actual values.
we say that a sybil strategy (G',U') for node i in G is successful, if some sybil has a better relative rank than i in G. Since imay use any of his sybils to perform transactions in the future, a sybil strategy is successful if at least one sybil has higher rank.
reasoning
Software and its engineering
train_127045
However, this category is not within the scope of this paper.
it will not be considered further.
reasoning
Software and its engineering
train_127046
The scenarios recovered are what we called design scenarios, so there are still some levels of abstraction to be traced up to requirements scenarios.
we believe our ideas can complement this work.
reasoning
Software and its engineering
train_127047
As prepare responses arrive, they are forwarded to all servers (line 228).
the threads from the prepare phase do not stop until all servers return responses or the commit phase completes.
reasoning
Software and its engineering
train_127048
Also, when an AD is a pointer to an MD in the MAR, the cor­responding MD in MAR is looked up, and the corresponding en­vironment variable is set appropriately.
the time-wise inte­gration of the FAR and MAR activities happens implicitly through the explicit pointers from the FAR to the appropriate entries in the MAR.
reasoning
Software and its engineering
train_127049
There is no semantic requirement to process requests in the order that they originate during the execution of an application.
the specialization controller could likewise traverse the entire queue, sort it into buckets referring to the same module and then make a choice which bucket appears to be the most promising in terms of expected return on investment.
reasoning
Software and its engineering
train_127050
v 3 For the area with size A, there are about A/( D2) 2 cells.
if these cells communicate simultaneously, the total throughput equals 2AR 2AP k2 v = v .
reasoning
Software and its engineering
train_127051
An attacker could exploit this vulnerability by setting the name parameter to ";alert( xss )//.
this would cause the resulting inline JavaScript to be the following, thus executing the attacker s JavaScript code: var username = "";alert( xss )//"; the code section of the application is dynami­cally generated with untrusted input and even with the code and data separated, there is still an XSS vulnerability.
reasoning
Software and its engineering
train_127052
The two lists are coordinated, so both ß are drawn if our red-black tree were a dictionary (with a key and value from the same type list ß ' .
for each pair of corre-for each conceptual node), separating the keys and the val­sponding nodes, the conditional types must evaluate to the ues could make functions accessing only the keys (e.g., is same branch.
reasoning
Software and its engineering
train_127053
While the deployment began with a few hundred users and has been augmented by early adopters, Rendezvous is beginning a phase of systematic deployments that will make it available to twenty to thirty thousand users over the next year.
while at present our results reflect the experience of volunteer "early adopters," we expect to be able to present a more general set of results in the future, at least for corporate conference callers (a not insignificant usage niche).
reasoning
Software and its engineering
train_127054
3.2.1 Taint Sources The goal of our algorithm is to (1) pinpoint the instructions that have a memory operand that needs to be symbolized and (2) pinpoint every global data location that stores an initial function pointer or a global data pointer which need to be symbolized.
our taint source contains the instruction address (PC) whenever an instruction involves a literal value operand, or a data value that appears to be a global pointer or a function pointer.
reasoning
Software and its engineering
train_127055
Strategies are a remarkably simple idea.
in the original formulation, a strategy is a function of type a -> () for some a: type Strategy a = a -> () a Strategy may evaluate its argument either in full or in part, and it may only return () (or diverge).
reasoning
Software and its engineering
train_127056
The replication management is considered to be an issue internal to the fragmented ob ject.
it is inherently replicated so that no external measures have to be taken to make the replica management fault tolerant.
reasoning
Software and its engineering
train_127057
The insiders are very skilled in the equipments or systems of the organization which they belong to and can delete immediately their traces after their improper activities.
it needs the acquisition of data in conformity with the leak type of confidential information for finding out the evidence.
reasoning
Software and its engineering
train_127058
For instance, for k = 300, at p1 =0.1, this guarantee applies for more than 90% of the acceptable defect rates; but at p1 =0.01, the guarantee applies form less than 3% of the acceptable defect rates.
the simple strategy can be applied only if a large uncertainty region is acceptable, which means that the Dis­patcher has a strong con.dence in the fact that the defect rate is actually null.
reasoning
Software and its engineering
train_127059
Java [14] and .NET [2], the value of f in the pristine state is assured to be an appropriately casted zero.
if a reference .eld in the pristine state is loaded into a root y , y will be directly overlooked by an immortal root.
reasoning
Software and its engineering
train_127060
However, we also do not want to impose upon the user the seemingly arbitrary cost of turning scalars into 1x1 matri­ces.
a scalar is implicitly promoted to be a 1x1 matrix in any context where a matrix is required.
reasoning
Software and its engineering
train_127061
Given this region, a reasonable guess as to the target host s location is at the region s centroid.
cBG uses the centroid of region R as a point estimate of the target s position.
reasoning
Software and its engineering
train_127062
We restricted the test bed to a LAN environment due to the fact that the JavaParty benchmark bundle has only about 100kB size.
the LAN scenario represents the worst-case scenario for OSGi4C becausetheloading timeis equalinaWAN set­ting(see Table2) butSOAP-basedWebservice invocation times increase.
reasoning
Software and its engineering
train_127063
There will also be a parallel thread with the execution time of (1 + Lfid - fi)Pi max and the relative deadline of (1 + Lfid)Pi max .
if we c si-1 2j m P 2j let Pi = 2 m i , the following inequalities are j=1 i satis.ed: (qi - 2)P max t i (1 + Lfid - fi)P max .i = + i (1 + fi)Pi max (1 + Lfid)Pi max ti.
reasoning
Software and its engineering
train_127064
As shown previously, it is not di.cult to see that the z-links used in this spanning tree are disjoint from other z-links used in case 1 and 2(b); thus all the links in this spanning tree are contention-free.
our algorithm for the construction of the spanning tree on a 3-D torus can be summarized as follows.
reasoning
Software and its engineering
train_127065
INTRODUCTION In the current globalized economic conjuncture of Brazil, the volume of negotiation carried on in the financial market has increased values and the competition, in this market, holds as a success differential the demand of information with more and more quality, transparency, quickness and low cost.
the pursuit for reduction in enterprise processes complexity using Information Technology (IT) is a challenge sought by software developers.
reasoning
Software and its engineering
train_127066
Between the trials, one question was the same.
the experiment contained 51 participants and four challenges.
reasoning
Software and its engineering
train_127067
As mentioned in Section 3.2, instance selection depends on the positions of the type parameters in the advice signature.
the mecha­nism cannot select an instance if all type parameters are not used in the signature of the instance advice.
reasoning
Software and its engineering
train_127068
Interactive Volume Segmentation Automatic volume segmentation is a dif.cult problem be­cause the image data is often noisy and the segmentation is dependent on the subjective interpretation of the ob­server [ONI05].
several interactive techniques have been developed to allow the user to specify input for semi-automatic segmentation.
reasoning
Software and its engineering
train_127069
So, a user has just to draw the data .ow graph of his application or follow a special wizard for complex applications.
sDAD gen­erates the (Application.xml, puts data and binary .les in a directory tree, and compresses all these data in a zip .le.
reasoning
Software and its engineering
train_127070
Mer3 has two precondition predecessors: {Mer4 }and {Mer5 }.
pCus2 = {{Cus3 }} and PMer3 = {{Mer4 }, {Mer5 }}.
reasoning
Software and its engineering
train_127071
As a result, they cannot be used, for example, to define the model of an arbitrarily­nested tree-widget such as a folder.
each of the approaches above makes a significant compromise.
reasoning
Software and its engineering
train_127072
Quality goals could di.er in di.erent quality levels, even can enter in con.icts.
self-management capabilities can enter in con.icts because of their contribution to quality goals in con.icts.
reasoning
Software and its engineering
train_127073
In order for the results from unsynchronized writes in the untrusted context to be seen in the trusted context, Codejail provides APIs which synchronize speci­.ed regions of memory in the untrusted context to the trusted one.3 In addition, system calls are restricted by a speci.ed policy when executing in the untrusted context.
codejail ensures that the untrusted library is sandboxed with respect to memory writes and system calls.
reasoning
Software and its engineering
train_127074
the concerns of version 1.0 of the EAM Pattern Catalog have been split up into prob­lems and forces, etc.
concerns are not explicitly covered in the rest of this article.
reasoning
Software and its engineering
train_127075
We do not display the input document as typed by the user, but an enriched version, to hightlight syntactic constructs.
we have to parse the input and then serialize the result.
reasoning
Software and its engineering
train_127076
Also transformations of basic decision models to feature diagrams are in general not always applicable, as feature modeling approaches in their very basic form usually oper­ates only of a subset of propositional logic [17].
this kind of feature diagrams can be regarded as intermedi­ate step.
reasoning
Software and its engineering
train_127077
Notice that it is di.cult for a node to trust the contents of a message, as it cannot determine whether that message originated at the ostensible sender s, or at a malicious device spoo.ng messages that claim to be from s. However, if a receiver does not receive a message in a round, it can be certain that no message was sent; the malicious nodes cannot forge silence.
we make frequent use of silence to authenticate data.
reasoning
Software and its engineering
train_127078
However, one well-known limitation of wrappers is that they do not preserve object identity.
code that employs == or instanceof on an expression of an interface type can get incorrect results if that expression evaluates to an expanded object.
reasoning
Software and its engineering
train_127079
Table 2 shows typical parameters of The e.ciency of garbage collection largely a.ects the per­formance of the system.
the block with the lowest garbage collection cost should be chosen as the victim block.
reasoning
Software and its engineering
train_127080
The constraint at s3 node is p and the constraint at s6 node is d, so the PCsets of these two paths respectively are as follows: PCset1 = {p, d} and PCset2= {.p, d}.
the DFC from s2 to s7 is the disjunctive of the two paths, DFC= (p.d) .
reasoning
Software and its engineering
train_127081
Each edge that is not used for the computation of a plane equation is forced to be located on it.
for each face Fi, one obtains deg(Fi)-2 equations, where deg(Fi) is the number of edges adjacent to Fi.
reasoning
Software and its engineering
train_127082
A more aggressive policy would require to com­pletely shut-off the card when no needed by any active ap­plication in the system.
more power can be saved, at the price of a larger wake-up delay needed by network re­association.
reasoning
Software and its engineering
train_127083
For a given analysis granularity, the capacity requirement ofa buffer at any point in time is .xed.
the X coordinate of any buffer in the two dimensional time-memory plane cannotbe modi.ed, i.e., the buffers can not be moved horizontally.
reasoning
Software and its engineering
train_127084
This implies the dominant thread must periodically check the request .ag.
the algorithm ensures starvation-freedom for the non-dominant threads only when the dominant thread checks the request .ag in.nitely often in any in.nite computation.
reasoning
Software and its engineering
train_127085
These barriers affect everybody, not only people with some kind of disability.
in order to avoid the barriers it is essential that both the content and the user agent be accessible.
reasoning
Software and its engineering
train_127086
Note that within the scope of class pattern $$C, the name $$C refers to exactly one element of the eventual set of matches for $$C.
the constructor pattern matches only one constructor, not a set (not that a class could have multiple nullary constructors anyway).
reasoning
Software and its engineering
train_127087
The CHECK-CAST intersection narrows the set down to {T2} (assuming T1 is not a subclass of T2), which is the resulting type set for field B.b.
any optimization can (optimistically) assume that a reference loaded from B.b must point to an object of type T2, as opposed to the closure of the declared type, T~.
reasoning
Software and its engineering
train_127088
Furthermore, the nodes did not live much longer than re­quested.
the applications were able to provide al­most optimal quality subject to the constraints present.
reasoning
Software and its engineering
train_127089
!k Lemma 2 implies xs = map (xs!!)
[0..length xs - 1] candidate (.)
reasoning
Software and its engineering
train_127090
Users find its interface easy to learn and use.
the goal of AgentWeb is to demonstrate that AgentSheets-like applications can be hosted entirely and natively on the Web.
reasoning
Software and its engineering
train_127091
However, there are many legitimate reasons why tests for databases and .le handling may not be independent.
it remains an open question to what extent our technique can be applied in such domains, and to what extent we need to extend its heuristics to reduce the number of false positives reported.
reasoning
Software and its engineering
train_127092
However, an important observation is that, since each operation crosscuts each visitor class, adding a new node to the existing class hierarchy will cause an invasive change to all of the visitor classes resulting in a maintenance nightmare.
no matter if it is a pure object-oriented or aspect-oriented implementation, the visitor pattern is applicable only under conditions when the node structure is static and does not change frequently.
reasoning
Software and its engineering
train_127093
So the last marked object in the path must be dirty, meaning that marking from live, dirty objects will lead to the object allocated during marking.
we could preserve correctness without allocating live during marking.
reasoning
Software and its engineering
train_127094
Thus, the upper bound on R, Rmax= 1,Vfinal1V=1,11(1+1).
since in this example, Vfinalis always smaller than or equal to V, the average value of Vfinalshould then be (1+11(1+1))12xVi the average rate of con.icting reservations, Ravg, equals to 1,Vfinal1V=1,(1+11(1+1))12= (1,11(1+1))12.
reasoning
Software and its engineering
train_127095
This incorrect con.guration makes IE Tab allow user names and password in URL address, which is blocked by IE [29].
iE Tab misuses WININET.dll w.r.t.
reasoning
Software and its engineering
train_127096
In Acme, a component may have a representation, which is a decomposition of that componentintoits constituentparts.
a componentmaybe a subsystem withits owninternal structure.
reasoning
Software and its engineering
train_127097
The library does not provide an interface ManagedRelationship; rather, we use the Set interface from the Java Collections Library.
staticManagedRel implements Set>, but StaticRel does not.
reasoning
Software and its engineering
train_127098
Besides,since.ash memoriesarepurely electronicdevices and have no moving parts, they have no seek or rotation latency like magnetic disks.
random access of .ash memory can be as fast as sequential access and the la­tency of .ash memory is almost linearlyproportional to the amount of data being accessed, no matter where the data Table 1: Magnetic Disk vs. NAND Flash [19] Read Write Erase Magnetic Disk 12.7 ms 13.7 ms N/A NAND Flash 80 µs (2 KB) 200 µs (2 KB) 1.5 ms (128 KB) is located.
reasoning
Software and its engineering
train_127099
Our experience has been that even with very careful scaffolding and practice exercises, the first few iterations seldom produce meaningful performance data.
we had to plan on many students needing to discard the data from the first two to three iterations.
reasoning
Software and its engineering