Sophisticated hypertext functionalities for Software Engineering

H yper text functionalities r epr es ent a for m of the distilled wisdom of the hyper m edia comm unity. G iven the peculiar natur e of the W or ld W ide W eb, it is very difficult to s ucces s fully propos e functionalities that becom e widely accepted. XM LC is a pr ototype of an XML br ows er that, given its m odular ar chitectur e and gener al s cope, can be pr oposed as the bas is for im plem enting s ophis ticated hyper text functionalities on the W eb. K EYW OR D S: X ML, dis plets , hyper text f unctionalities INTRODUCTION The community of hyper text f unctionalities w as born in or der to identif y and list the f unctionalities intr ins ic to the idea of hyper text, and to either verif y them or intr oduce them in other communities s uch as document management s ys tems , the Wor ld Wide Web, sof tw ar e engineer ing [ 14] , etc. I n par ticular, the Wor ld Wide Web has developed according to w ays that w er e ver y peculiar and diff icult to pr edict. For ins tance, the WWW community valued the development of s tandar ds and pr otocols more than f unctionalities . This lead to the cr eation of s ome dozens of dif f er ent languages and pr otocols that ar e necess ar y to mas ter the tas k of cr eating s atis f actor y Web s ites . I n our opinion this r ichnes s of languages show s on the one hand, that ther e exis t the pos s ibility of implementing a large number of inter esting f unctionalities, and on the other hand, that unf ortunately the WWW does not enf or ce or even f acilitate them, s o that their us e depends on the w ill and aw ar eness of the authors of Web pages and sites . F urther mor e, this r ichnes s of poss ibilities is coming to the detr iment of s implicity, w hich was once the r eal advantage of the Wor ld Wide Web over other sys tems s uch as G opher or F TP . I n the near f utur e, and jus t to mention nonpr opr ietar y ef f or ts , any Web author w ill have to deal w ith at least ten diff er ent and non-tr ivial languages or pr otocols, namely HTM L, H TTP , CS S , ECMA s cr ipt ( or any of its pr opr ietar y dialects , s uch as J avas cr ipt or J s cr ipt), X M L, X S L, X P ath, X Link, XM LS chema, RD F, and WebD A V , plus many others that at the moment ar e s tar ting to catch on. S o, w hile ther e r eally exis t the pos s ibility, in the languages , to provide s ophis ticated hypertext f unctionalities, w e have to w ait f or Web applications to actually provide them in a us able w ay. Y et, the X M L f amily is a cons iderable advancement over pr evious languages and s tandar ds . The poss ibility given by X ML [ 4] to def ine a s yntax ( i.e., a Document Type D ef inition, or D TD ) tailor ed f or one’s document class es , and to us e s tandar d X M L tools to create, ver if y and exchange data is a r eal bonus . But in our opinion the s trength of XM L lies beyond the capabilities to def ine communitys pecif ic D TD s: f or ins tance, it is becoming convenient to us e it even f or applicationonly data, that is , f or objects that ar e not natur ally meant to be dis played to a human us er . A dditionally, X S L [ 10] pr ovides much to X M L in ter ms of r each and f lexibility. X S L includes a mapping language [ 9] that can be us ed to tr ans for m an X M L document into another one. Cur r ently its most impor tant us e is to tr ans f orm an X M L document into a f or mat that can be dis played by a br ow s er : thus f or ins tance Micr osof t Inter net Explor er 5 can accept X M L documents of any D TD and us e X S L to trans f or m them into an H TM L document that can then be pr oper ly dis played on a computer s cr een. O ur longter m pur pos e is to cr eate an environment that, w hile relying on s everal exis ting Web languages and pr otocols, can pr ovide fundamental hyper text f unctionalities in a s treamlined and eas y w ay. I n this paper w e will concentr ate on br ow s ing and displaying hyper text data. O ur approach is par ticular ly us eful to make s of tw ar e engineer ing envir onments " WWW-aw ar e" : the documents of the sof tw ar e proces s tend to be compos ed of sever al diff er ent chunks , s ome of text, s ome of f or mulas in s pecial notations , and s ome of s tr uctur ed gr aphical diagr ams . Cur r ently it is ver y diff icult to tur n thes e documents into pages that can be made available thr ough a Web brow s er, s ince each f ormula and each diagr am need to be conver ted into a pas s ive image. I n pas t papers w e dis cus s ed displets ( [7] and [ 8]) , our pr opos al to pr ovide f lexible s uppor t for s pecial r ender ing needs that author s may have. D is plets ar e sof tw ar e modules ( curr ently they ar e J ava clas s es ) that ar e as s ociated to each element in an X ML document and that pr ovide s ome r ender ing behavior f or that element. Support f or the mos t common element types is pr ovided ( for ins tance, text elements and par agr aphs) , but it is pos s ible at any time to add new modules enabling s pecialized r ender ing s emantics f or s pecif ic needs. D isplets pr ovide a w ay ar ound the af or ementioned limitations , allowing the s emanticallyr ich des cr iption of f or mulas and diagr ams to be expr ess ed in X M L and to be dis played in the brow s er in their f ull gr aphical r ender ing. A dditionally, the pos s ibility of activating in some ways the diagr ams , ver if ying their cor r ectnes s and pr oviding multiple view s s eem impor tant and eas ily pos s ible in a gener al w ay with the displet appr oach. I n this paper w e dis cuss X M LC, a r ecent implementation of our dis plet ar chitectur e ( als o br ief ly intr oduced in [ 3] ). X M LC can be cons ider ed as a ver y general architectur e f or pr oviding ver y s ophis ticated f unctionalities to documents cr eated in the X M L for mat. While the over all des ign goals ar e to cr eate a complete author ing envir onment for s ophis ticated hyper media, in the cur rent paper w e concentrate on s ophis ticated br ows ing of X ML data. I ndeed, the ar chitectur e des cr ibed her e can be f r uitf ully used f or more than visualization, f or it is an extr emely general w ay to as s ociate behaviors to X M L elements , and thus to produce active documents that per f or m computations , enact goals , pr oduce r es ults . We call thes e documents declar atively active docum ents ( or D A Ds ) because of this char acter is tic. This paper is s tr uctur ed as f ollow s : in the next s ection we discus s s ome of the most impor tant hyper text f unctionalities on the Web. Then w e clas s ify s ome hyper text f unctionalities that are natur al to s ingle out in nor mal s of tw ar e engineer ing documents . N ext w e pr ovide a f ew backgr ound infor mation about X M L and r elated notew or thy s tandar ds . I n the f ollow ing s ection w e dis cus s the curr ent ar chitectur e of X M LC, and pr ovide examples of s ome of the dis plet class es we have cr eated. O f par ticular impor tance in our view ar e the packages f or dis playing notations r elevant to s of tw are engineering, w hich have constituted f or s ever al r eas ons our main target for the implementation of displets. 2 HYPERTEXT FUNCTIONALITIES ON THE WEB I n [ 2] a lis t of 9 f undamental ( accor ding to the author s ' opinions ) hypermedia f unctionalities w ere pr opos ed and dis cus s ed, w ith the unders tanding that few of them, if any at all, w ere either available on the Wor ld Wide Web or exploited to their f ull potential: • Typed nodes and links • Link attr ibutes and s tructurebased quer ies • Tr ans clus ions, w arm and hot links • A nnotations and public vs . pr ivate links • Computed per sonalized links • Exter nal link databas es and link update mechanisms • G lobal and local over view s • Tr ails and guided tour s • Backtr acking and his torybased navigation. Thes e items wer e selected f rom a longer list of 25 items as s embled at the 2nd H TF wor ks hop in conjunction with the H yper text ' 96 conf er ence [ 1] . A t the moment, pr obably, all of thes e f unctionalities could be eas ily implemented on the WWW. S erver side CG I , s ervlets and D BM S applications , as w ell as clients ide plugins, J ava and J avas cr ipt pr ogr ams allow now a degr ee of f reedom in cus tomizing the WWW unpr ecedented in any other hyper media s ys tem ( even those that did implement s ome of thes e f unctionalities) . The r esear ch and commer cial communities have in f act alr eady explor ed s ome of thes e functionalities in the last f ew year s . Yet, few of them have r eally caught on w ith the larger WWW community, or even f ound a s mall vis ibility stand-point thr ough the available commer cial applications . I t is indeed our opinion that no J ava applet, CGI application or other cus tom concoction can pos s ibly pr oduce any relevant change in the w ay the WWW is us ed. The r eas on for this is that these w ould all be added f unctionalities to the cor e sets in s erver s and br ow s er s , and, unf or tunately, the WWW in neither the set of s er ver f unctionalities , nor the s et of br ow ser f unctionalities . The f act that the WWW is not a s ys tem, or a s et of inter dependent s ys tems, but a s et of pr otocol and languages , is obvious yet not s uff iciently unders tood. N o s ingle s ystem can provide added value to the WWW as a w hole. A lmos t no organization ( in many cas es not even M icr os of t or N ets cape) can intr oduce a new f unctionality in its pr oducts and f ind out that the WWW as a whole catches on. The WWW mus t not be impr oved in the sys tems , but in the way it actually w or ks: by changing the under lying languages and pr otocols ( H TM L, HTTP , CG I , etc.). We can gr oup the above-mentioned f unctionalities in tw o lar ger f amilies : thos e that add to the active participation of the user s in the pr oduction of inf ormation, and thos e that add to the exploration of the available infor mation. O n the one hand, annotations , pr ivate links and computed pers onalized links ( that r equir e exter nal link bas es and link update mechanisms to w or k on a lar ge s cale) allow f or the active par ticipation of r eader s to the nodes they r ead. O n the other hand, over views , trails , guided tour s and s ophis ticated backtr acking patterns ( that require r icher types and attr ibutes f or nodes and links ) enhance the navigation and the access to the inf or mation of the hyper base. F inally tr ans clusions and links of var ious temper atur e pr ovide both a r icher expr es s ive means for author s, and a r icher explor ation means f or readers . Both f amilies s hare the s ame pr oblem: they ar e not f unctionalities that can be exper ienced by the s ingle us er , i.e., that one enlightened us er can adopt f or his /her ow n pur pos es and be enr iched by us ing them: they ar e f unctionalities that have to be s har ed by a large community in or der f or them to f ully provide their benef its: ther e is little point in us ing an exter nal link databas e, if w e can't s har e our links w ith our colleagues ; ther e is little point in annotating or trans cluding, if w e can' t publish our notes and tr ans clus ions; ther e is little point in being able to cr eate over view s and guided tour s on s ome collection of documents, if w e can' t publish them f or our r eader s . Thus thes e f unctionalities must be dictated thr ough the s tandar ds and pr otocols that make up the Web, r ather than thr ough any s pecif ic application. M ore r ecently, in [ 16] f our hypermedia f unctionalities w ere f urther identif ied: • editable br ows er s • s tor ing document content and link anchor s separ ately • exter nal linkbas es • displaying link s pans , node and link attr ibutes I n all thes e cas es, actual WWW protocols w er e cited that could pr ovide the neces s ary expr es s ive power to implement these f unctionalities : WebD A V [12] pr ovides clients w ith r emote w riting pow er , thus make editable br ows er s a r eal poss ibility. X P ointer [ 11] and X Link [ 13] allow exter nal links , thus making it poss ible to separ ate content and link, and to put links into exter nal linkbas es . RDF [ 5] allow s ar bitrar y metainf or mation to be added to any Web document, and to be us ed for clas s ification, indexing, and s earches . A shar eable long ter m goal is to identif y a s ingle, simple and s tr eamlined ar chitectur e to provide all thes e f unctionalities us ing WWW pr otocols and hiding the complexities behind the protocols used. With X M LC, w hich will be descr ibed in the next section, w e ar e pr oviding a s ingle, eas y to use and easy to expand ar chitectur e f or br ow s ing X ML documents . We consider it a f ir s t s tep in that dir ection. HYPERTEXT FUNCTIONALITIES IN A SOFTWARE DEVELOPMENT PROJECT A s oftwar e developm ent proces s is the des cription of both the activities and the documents to be pr oduced dur ing the development of a sof tw ar e product. I n s hor t, a sof tw ar e proces s is a method to pr oduce a s oftw ar e pr oduct; it pres cr ibes in details all the documents to be pr oduced and all techniques and tools to be us ed in all development phas es s tar ting fr om the explor ation of the concept of a new pr oduct and ending w hen the pr oduct is f inally r etir ed f r om operation. I n the las t year s , s of tw ar e engineer ing envir onments have evolved mainly in the explicit s uppor t they of fer to s pecif ic s of twar e pr ocess models . This means that mos t r es ear ch ef f orts have f ocus s ed on how a sof tw ar e proces s is des cr ibed and how its activities ar e contr olled and enacted by a «pr oces s engine», namely a pr ocess center ed pr ogr amming envir onment that divides the w hole s of tw are engineering lif ecycle in s ever al s ubs equent and/or par allel phas es , each w ith its ow n char acter is tic input and output documents . The s et of documents r elated to a s of tw ar e development pr oces s ar e many, of dif f er ent f or ms , f or mats , and s pecif icity. Documents belonging to diff er ent phas es of the s of tw ar e pr oces s w ill s ometimes have the s ame topic, but w ith a dif f erent level of detail and gr anular ity. D ocuments belonging to the same phas e w ill be dif fer entiated in pur pose, s tr uctur e and content. S ingle documents belonging to one phas e may contain par ts of dif f ering str uctur e, notation, and pur pose: f or instance, the s ame topic may be handled w ith a f r ee-text des cription, a f or mal s pecif ication us ing an appr opr iate mathematical notation, a gr aphic depiction of its par ts , etc. A key iss ue that has become per vas ive and obvious in r ecent year s is the s uppor t f or hyper text f unctionalities in the data f or mat. I n its s imples t for m, hypertext is the pr ovis ion of connections and r elations hips between documents and text chunks , allow ing r eader s to move f r om a document to another one (navigation) following a dif fer ent or der than the one imposed by the linear s equences of the documents thems elves . I n a complex s ituation s uch as the one enacted in the s oftw ar e development pr ocess , the r elations hips could be clas sif ied in scope as f ollow s: ⇒ inter phas e relations hips , or the r elationships exis ting among documents belonging to dif f er ent phas es of the s of tw ar e pr oces s . ⇒ intr aphas e relations hips , or the r elationships exis ting among dif f er ent documents of the s ame pr oces s phas e. ⇒ inter par t r elations hips , or the r elations hips exis ting, within a s ingle document, betw een dif f er ent parts and poss ibly dif fer ent notations ( f r eetext, s tr uctur ed text, mathematical notations , gr aphical depictions . ⇒ intr apar t r elations hips , or the r elations hips exis ting w ithin a s ingle par t or notation of a document, betw een the atomic entities that compos e it ( f or ins tance, between dif f erent elements of the s ame s chematics ) . A mor e compr ehens ive and gener al def inition of hyper text is “ r elations hip m anagem ent for databased applications”. A n application in our s ens e is not s imply a pr ogr am, but a str uctur ed s et of pr ocedures that are concer ned w ith a collection of data elements and inter es t computer pr ogr ams , people, pr oces s es , plus w hatever s tor age, retr ieval, tr ansmis s ion and pr oces s ing engines ar e us ed f or their management. The same application can be used w ith dif f er ent data, and the s ame data can under go dif f er ent applications . So, accor ding to our def inition, hyper media is all that is concer ned w ith s tructuring and giving access to the par ts compos ing an application thr ough their inter r elations hips. Thes e r elations hips do not simply have to be among the data elements, or be compos ed of s pecif ic, ad hoc r ef er ences betw een tw o specific objects. We take a br oader view of the idea, der iving f r om [ BV97] a lis t of clas s es of r elations hips . Schem a r elations hips: the connections cr eated by the s tr uctur ing of the data elements in s epar ate objects w her e the r elation is appar ent in the s tr uctur e its elf . The s chema r elations hips deal w ith the s tructures of the entities of the s of tw ar e pr oces s : the modules of sof tw ar e packages , the f unction points of a complex pr ocedure, the elements of a complex data type, the methods of an object, etc. M odular des ign pr actices ( topdown, bottom-up, str uctur ed, object oriented, etc.) imply that s ome entities ar e def ined once and r ecur in s ever al dif f erent places in the s ame document or acr oss documents . O ntological r elations hips: the connections linking data elements , progr ams , people, pr oces s steps or under lying w or king envir onments with the par ameter s and des cr iptive inf ormation that accompany and def ine them. Ontological r elationships pr ovide gener ic inf or mation about the individual entities of a document: the meaning of an acronym or of a s pecialized ter m, the numerical values of a quantitative des ign cons tr aint, the global identif ier of a document s ection, is s ue, r equir ement or entity, etc. S uch infor mation chunks ar e not neces sar ily contained in any other document and can us ually be deduced by the definition of the element, by its s pecif ication, or by an apposite data dictionar y pr ovided f or this pur pos e. O ccur r ence r elations hips: the connections betw een all appear ances and uses of a data element, pr ogr am, per son, etc., in the application. s ometimes the same entity, r equir ement, or f unction appear s in s everal documents , for dif f er ent pur pos es and in dif f erent detail. O ccur r ence r elations hips exis ts betw een an entity and all the places in the document w her e the inf ormation is pr es ented, dis cus s ed or detailed. I ndexes s pecif y all the occur r ence of a given ter m. Specialized indexes may create a list of all the occurr ences of only thos e ter ms that ar e deemed inter esting. Tables of content, on the other hand, provide a gener al s tr uctur e of the def inition of terms . By combining thes e two s ervices, one can easily pr ovide all the occurr ence r elationships in documents . Pr oces s r elations hips: the connections betw een a data element, pr ogr am, per s on, etc. and the application' s pr oces s es that can or do inter act w ith it. We s ay that there is a pr oces s r elations hip betw een an entity and all the tasks in the s of twar e pr ocess es that deal with the entity. F or ins tance, ther e is a proces s r elationship betw een dr af ts of the s ame document as it pr ogr es s thr ough all the s tages of cr eation, verif ication and modif ication. O r, a pr ocess r elations hip exis ts connecting the test s eries and their output, becaus e each output is the res ult of a pr oces s on a specific tes t s uite. Structural r elations hips: the connections that embody a str uctur e of cons tr aints and r ef er ences . S ome infor mation chunks may be s een as a dif f er ent view of s ome data items that may or may not be s how n autonomous ly. F or ins tance, if a r ead-only method of a clas s is a s imple computation based on other , r ead/w r ite elements , a s tr uctur al r elationship exis t betw een the method and the f ormula. M or e abstr actly, the choice of a given s of tw ar e pr oces s model will impos e the cr eation of a ser ies of given documents , w hos e str uctur e and layout is often given in advance. Thus ther e exis t a s tr uctur al r elations hip betw een each s tr uctur e and the par t of the s pecif ication of the sof tw ar e model that r equir es it. D ynam ic r elations hips: connections that ar e not intr ins ic to the application model or in any other w ay know n in advance, but that become appar ent dur ing the lif e of the application. These r elations hips may der ive f rom applying s ome logic and computation to the s tate of the document, or may der ive f r om exter nal events that happened s ince the document w as w r itten. A dynamic relations hip thus is a datamining dis cover y about the us age, s tr uctur e, occur r ence, etc. of entities and documents . G ener ally, thes e relations hips pres ent thems elves during the lif ecycle of a document, as oppos ed to the time of its creation. That is to say, they cannot be dis cover ed in advance, during the des ign of a document, but ar e the res ult of computations on the ins tances of documents . I n the cas e of the s oftw ar e pr oces s , the mos t impor tant computations that can be per f ormed on documents is validation and ver if ication of their cor r ectnes s , completenes s and cons is tency. Whenever a pr oblem is f ound out, either automatically or by a human r eader , a r elations hip is created on the r elevant documents . Besides actual pr oblems, dynamic r elations hips automatically dis cover ed by appr opr iate engines may impr ove the cons istency in ter ms , inter f aces , libr ar ies , etc., w hich may be extr emely impor tant f or large and complex s of twar e pr ojects . Ad hoc r elations hips: Ad hoc r elationships ar e all the r elations hips w hose exis tence cannot be deter mined by a r ule, but are cr eated becaus e of an ad hoc decis ion ( usually by a human, but not neces s arily) . These ar e the w ell-know n hyper text links in the str icter s ens e. F or ins tance, this includes all connections betw een s tr uctur ed elements of the documents and non s tr uctur ed chunks (s uch as annotations , comments , dis cus s ions , etc.) . F ur ther more, of cours e, all links and r ef er ences to documents not pr oduced w ithin the s of tw ar e pr oces s are ad hoc. F ur ther more, w e include in this categor y all connections that, f or s ever al ad hoc r eas ons , cannot be included in the other categories ( s uch as links to unclas s if ied bug r epor ts , unimplemented cons tr aints , deliberate violations of r equir ements and specifications , etc.). Inter-phase relationships Intra-phase relationships Inter-part relationships Intra-part relationships Schema relationships Recurring definitions Use-definition Use-definition Not appropriate Ontological relationships Methodology explanation Terms and objects dictionaries Terms and objects dictionaries Object properties Occurrence relationships Table of contents and indexes Table of contents and indexes Table of contents and indexes Table of contents and indexes Process relationships Not appropriate Test sets and results Process descriptions Process descriptions Structural relationships Stub generation Not appropriate Not appropriate Computed class members Dynamic relationships Inconsistency reports Inconsistency reports Inconsistency reports Inconsistency reports Ad hoc relationships Comments and references to ext. literature Comments and references to ext. literature Comments and references to ext. literature Comments and references to ext. literature Table 1. Relations hips among sof tw ar e proces s documents To ever y type of relations hip corr es ponds a type of link. With the exception of ad hoc links , which have to be created one by one by s killed people, and dynam ic links, which s hould be cr eated dur ing the lif etime of the document bas e by ad hoc analys is and data mining applications , all other types of r elations hips can and should be made available to us er s w ithout s pecif ic human inter vention. Thes e tw o class if ications , in s cope and types , provides us w ith the gr id in Tab. 1, w hich w e filled w ith the types of r elationships that ar e relevant f or the s of tw ar e pr oces s . XML AND RELATED STANDARDS F ormed in 1994 by the special inter es t gr oup on S G M L of the World Wide Web Committee, the X M L w or king gr oup s et to w ork to pr ovide an S G M Lbas ed gener alized mar kup language that could pr ovide mos t of the f unctionalities of S G M L, w hile r etaining the s ame s implicity of H TM L f or the casual author and the developer of tools . The w or king gr oup divided the ef f or t in sever al independent but r elated languages that cooperate in or der to pr ovide a complete and sophisticated mar kup envir onment. F or our pur pos es , the most impor tant pr opos als of the XM L f amily ar e the X ML gener alized mar kup language its elf , the X S L s tyles heet language, and the D O M document object model, which w ill be dis cus sed s ingular ly in the f ollow ing s ections . X M L 1.0 ( Extens ible M arkup Language, [4] ) is the f ir st and mos t s table propos al of the XM L f amily. I t is a pr oper s ubs et of S G ML, r ather than an application of it ( like H TML) . XM L is a gener alized markup language, wher e author s can decide the s et of tags they ar e going to use in w r iting their documents , and it is used at its bes t to expr es s the s tr uctur e of the document and the semantically r elevant elements , r ather than their typographical proper ties . J ust like S G ML documents , X ML documents lack any machineinter pr etable s emantics . That is , X M L dictates about the s yntax of the mar kup, not about its meaning. This means on the one hand that author s ar e fr ee to as sign whatever meaning they w ant to the elements they choos e to employ, but als o that ther e mus t be a w ay f or pr ogr ams to make use of thes e elements in an appropr iate w ay. The s olution taken w ithin the X M L f amily is to m ap ( or r ewr ite) the s our ce X M L document ( containing elements meaningf ul to the author ) into a dif fer ent X M L document containing elements meaningful f or the pr ogr am that has to per for m the application. Thus, a document containing elements s uch as s ection, titles, f ormulas, s entences , etc., all meaningf ul to its human author , w ill be r ew r itten into a new document containing elements s uch as blocks , par agr aphs , inline s tyles , w hite r egions , and s o on, all meaningf ul to a dis playing or printing pr ogr am. This is done by X SLT [ 9] , the tr ans f ormation part of X S L (Extens ible S tyles heet Language) . X SLT is a mapping language f or X M L elements . Each XS L s tyles heet is made of r ules compos ed of a patter n and an action. The patter n identif ies an element of the X ML s our ce to w hich the action should apply; a pattern may s pecif y all elements of a given type, all elements contained in a given s ubtr ee, all elements having a given attribute, etc. When the mos t appropr iate patter n is found f or the cur r ent X M L element, the action par t is cons ider ed. X S L actions are s imply XM L s ubtr ees that ar e w r itten in the des tination document in place of the element being cons idered. The D ocument O bject M odel [ 17] is a platf ormand languageneutr al inter face that allow s pr ogr ams and s cr ipts to dynamically acces s and update the content, s tr uctur e and style of documents . The D ocument O bject M odel pr ovides a s tandar d set of objects for r epr es enting H TM L and X M L documents , a s tandar d model of how thes e objects can be combined, and a standar d inter f ace f or acces s ing and manipulating them. Thr ough D OM it is pos s ible to develop applications that us e X M L documents in a very gener al w ay. THE ARCHITECTURE OF XMLC X M LC ( X ML Compiler) is an architectur e f or r ender ing dis plets . X M LC r elies upon technologies and languages s uch as XM L, XS L and D O M.