SV-EC Ballot Resolution Committee Meeting Monday May 9th 2005, 11:00am - 1:00pm PST [Minutes distributed for approval at next (sv-ec) committee meeting] (01111222200) Day (41258158949) (00000000000) Month (44444444455) (00000000000) Year (55555555555) --------- Attendees ---------- (-AAAAAAAAAA) Arturo Salz (AAAAAAAA-AA) Brad Pierce (AA--AAAAAAA) Cliff Cummings (AAAAAAAAAAA) Dave Rich (AAAAAAAAAAA) Francoise Martinolle (A-----AAAAA) Karen Pieper (AAAAAAAA-AA) Mehdi Mohtashemi (AAAAAAAAAAA) Neil Korpusik (AAAAAAAAAAA) Ray Ryan (AAAAAAA-A-A) Steven Sharp (AA-AAA---AA) Surrendra Dudani (-AAAAAAAA-A) Gordon Vreugdenhil (-------AAAA) Phil Moorby (----------A) Stu Sutherland (-----A-A---) Dennis Brophy (---------A-) Don Mills (------A----) Tom Fitzpatrick ^ ^ ^---- Non-Voting meeting (4/29/2005) ** Minutes taken by Neil Korpusik and Mehdi Mohtashemi Agenda: 1. IEEE patent policy ref: http://standards.ieee.org/board/pat/pat-slideset.ppt Move: Assume it was read - Cliff Second: Neil Abstain: Opposed: 2. Review Meeting minutes, May 4th, 2005 http://www.eda.org/sv-ec/Minutes.html http://www.eda.org/sv-ec/Minutes/SV-EC_BallotRes_Meeting_May_4_2005_Minutes.txt Move: Cliff Second: Arturo Abstain: Gordon (not there), Steven, stu Opposed: approved ** Champions meeting Today at 1pm. BC and CC have until tomorrow to deal with any issues that come out of it. ** P1800 meetings SVDB items for review and closure The mantis items have been checked ACKNOWLEDGED. Karen has requested the following: Please have someone in your committee review each issue and either close the issue or 2) add a bug note indicating the trouble with the editing, reassign the issue to the SV-LRM and send a note to Stu and I indicating that this was done. AI: all members of sv-ec committee to review each Mantis items which has been approved/closed. 3. SVDB items for review and closure There were several Mantis items which Neil and I thought we can bring to the closure today before the champion's meeting. Here is a list we can start to review. ballot item 7- Mantis 344 http://www.eda.org/svdb/bug_view_page.php?bug_id=0000344 Arturo: modifying 17.1, includes the BNF change, simplified Dave's proposal. Move: Dave to accept 344 Second: Arturo abstain: opposed: b) Mantis 246 - random object stability example http://www.eda.org/svdb/bug_view_page.php?bug_id=0000246 (Arturo mistakenly closed the errata, reopened by Karen) Arturo: object stability, stability on object not on types or semantics. MOVE: Neil to close 246 as Not-a-BUG Second: Arturo abstain: opposed: 246 CLOSED NOBUG c) Mantis 384 - duplicate of 514 (approved) mailbox type http://www.eda.org/svdb/bug_view_page.php?bug_id=0000384 (in 384, it indicates duplicate of 517, it should indicate duplicate of 514 --- make the correction and close 384) Mantis 384 ) [approved on 4/25/05] (duplicate of 514) MOVE: Neil, to close 384 as a duplicate of 514 Second: Cliff abstain: opposed: d) need to close this 248 - missing ; in definition [has been fixed] http://www.eda.org/svdb/bug_view_page.php?bug_id=0000248 MOVE: Karen moves that 248 be closed as not a bug Second: Brad abstain: opposed: e) 171 - configuration of programs http://www.eda.org/svdb/bug_view_page.php?bug_id=0000171 Steven: Have those changes been made. has sv been updated to allow program blocks in configurations? Gordon: does it boil down to cell identifier in the grammer, in 2005 draft, extremely general 13.6 in 1364 vpi cell is name of cell bound to the module instance. Steven: does it use vague terminology, design unit. Arturo: informative clause was inconsistent with the 23.2 p1800, describes cells in libraries MOVE: Artuo Second: Gordon abstain: opposed: 171 approved/resolved f) Mantis 725 Matching of class specializations http://www.eda.org/svdb/bug_view_page.php?bug_id=0000725 (Addresses lack of explicit rules for the matching of class specializations) Gordon: intent section 23, class parametrization works close to c++ templitization, verilog elaboration model does not correspond to c++. What constitutue the specialization match and the domain of specialization. Arturo: verilog instantiation of real object as in module, here you create a concrete type when you specialize the object. When do two generic classes represent the same type. Cliff: we can do name to parameter passing. Gordon: nothing new in terms of how you use it, how you identify. Steven: both type and parameter are of the same, Gordon: narrow rule, type and the value to treat them MOVE: Gordon to accept 725 Second: Arturo abstain: opposed: Brad: the second proposal (May 7th, 2005). the bnf proposoal allows #() to specify that all of the defaults are being used. It would highlight class being specialized or not. Gordon: the example, myvector is an actual class specialized with default value of 1. the bnf would allow to make it more explicit, user clue to make him/her remember that this is specialization. Steven: if you do specialliztion and pass exact values with default, is it different that default one. what if the default matches someone elses explicit specialization?. Gordon: if it matches the default you get something that matches the default. MOVE: Karen moves to accept combined/merged proposal for 725. Second: Gordon abstain: opposed: AI: Arturo update proposals into combined/merged one for 725 g) Doug's manits entries: (I created new version of proposal for 724 & 715 as an example) 724 - trans_range_list BNF and example http://www.eda.org/svdb/bug_view_page.php?bug_id=0000724 Ray: single item was implied in the examples, used to have at least 2 items in trans_range_list Move: Ray to accept 724 Second: Cliff Abstain: Opposed: 715 - functional coverage examples http://www.eda.org/svdb/bug_view_page.php?bug_id=0000715 Ray: it does not absolutely require the (), it did not mean what you thought, the example means something different without the parens The proposal is to add in the parens for clarification. Move: Ray move to accept 715 Second: Cliff Abstain: Opposed: 720 - coverpoint syntax http://www.eda.org/svdb/bug_view_page.php?bug_id=0000720 Ray: in the example it looked like $ used for range, it replaces range-list with open_range list which means $ in the list. Move: Ray moves to accept 720 Second: Gordon Abstain: Opposed: 714 - cross example http://www.eda.org/svdb/bug_view_page.php?bug_id=0000714 Ray: the intent was the coverpoint, if variable a, b. Note - Some people were not able to see the colors in the proposal. a,b are in blue (v_a,v_b are in red and struck out). Stu: it helps when the proposals are in pdf format. Surrendra: the text under the example also needs to be updated. Move: Ray Moves to accept 714 with friendly amendment above, to fix the text under the example Second: Arturo Abstain: Opposed: AI: Ray to modify the proposal to make corrections for a,b page 334, and the friendly amendment. 703 - pre_randomize() and post_randomize() http://www.eda.org/svdb/bug_view_page.php?bug_id=0000703 Ray: built-in to prototype, super is not an identifier by itself, Gordon: super class is not an item. Arturo: this is a psuedo code, but the text explains it. Stu: move second if to the front of the sentence. --> If the class is a derived class and no user ... Mehdi: Needs to be done in both paragraphs. **** Friendly Amendment: if the class is not a base class is removed from both paragraphs, and If the class is derived class and is added to beginning of the line (second sentence in each of the two pragraphs starting with Users...) Move: Ray accept 703 with friendly amendment Second: Arturo Abstain: Opposed: AI: Ray will update the proposal with the friendly ammendment and place pdf file. h) Ballot item 95 Mantis 512 - wordsmith of object copy section 12.25 bullet 2) http://www.eda.org/svdb/bug_view_page.php?bug_id=0000512 Francoise: copying of object, derived class can be assigned to super-class, but not the other way around. Arturo: a super class is the same type, in the different heriarchy it would be the same type. Gordon: it would not be the same type under matching rule, Arturo: derived to base is assignment compatible Brad: is it about assignment compatibility not about type matching. Francoise: assignment of compatibility of object is not well defined now. Brad: would you call that implicit cast, same heirarchy, but not the same type. Arturo: you need an explicit cast. Ray: does this need to refer to object handle, not the real object. Francoise: related to matching specialized classes, Gordon: no, it is the same type, we do not need to, no reason to 12.25, copying is odd, you are not talking about defining copy method. Ray: this sentence in the paragraph does not belong in here Arturo: the assignment compatilitiy is tronger than struct. Gordon: this is talking about copying object. Brad: unpacked struct is very restrictive. Gordon: Would like to strike point 2 from 12.25, and change word four to three in the previous paragraph. Arturo: the whole section was meant to be informative Motion: to strike bullet 2 in 12.25 striking restructring item 1,3,4 to 1, 2, 3. change four in paragraph to three MOVe: Gordon Second: Brad Abstain: Opposed: AI: Brad - place the proposal to the mantis database. i) 236 - zero valued cycle delay http://www.eda.org/svdb/bug_view_page.php?bug_id=0000236 Steve - to be consistent it should not wait at all. Art - the proposal is ill-defined. Gord - the clocking event is immaterial here. - ##0 should be a no-op in a clocking block. Decision - leave 236 open. Mehdi: we do not need to vote on it, remains open as an item for future. AI: Gordon. add a bugnote, to effect that this was reviewed and discussed j) 211 - built-in scope randomize http://www.eda.org/svdb/bug_view_page.php?bug_id=0000211 Brad: I do not think this is a simple issue, like to put it off for now. LEFT for future review **** item 101 mantis 523: queue return type Brad: altered the content of the queue. I would prefer we get rid of it. not there on all of them. Steven: that representation matters, if there is header that points to the queue, everybody sees the change. question about pass by ref - is that correct? Mehdi: what about other methods? - 2.6 and 2.7 also apply Brad: is it true that these methods have refports that pass in the queue. also 2.6, 2.7 Ray: are these methods members of the queue type, Arturo: usually we don't show the this pointer so we don't really need the ref here. - it looks like there was an editor's mistake here. The original proposal did not contain this arguments. Brad: howabout to get rid of ref ports all together, and 2.6, 2.7 Friendly amendment: 1) For methods: insert(), push_front() and push_back(), remove the ref ports in the prototype 2) do not add any thing to the delete() prototype. 3) q type return value is changed to element queue_t MOVe: Brad to accept Mantis 523 with above friendly amendment Second: Francoise Abstain: Opposed: *** Mantis 729 bnf for program_generate_item Brad: no path to generate through program in BNF, MOVE: Brad moves to accept 729 Second: Gordon Abstain: Opposed: *** 340 is closed as a duplicate. of 523 MOVE: Arturo Second: Karen Abstain: Opposed: *** Mantis 532 fixing syntax boxes Brad: Purely editorial MOVE: Brad Second: Cliff Abstain: Opposed: *** Mantis 730 4.7.3 putc return type Arturo: putc gets a byte, but getc has not been fixed, in c you also pass int, Steven: c upconverts all arguement, Arturo: we modified putc to accept byte only, did not modify to make it symmetrical. getc in a concatenation you get 32bit instead of 8bit, if you do file operation you can return a empty. Ray: are we talking about 4.7.3, not to do about i/o, statement here says it returns 0. Motion: to modify 4.7.3, to change the return type for getc to byte Move: Arturo Second: Neil abstain: Opposed: AI: Brad place the proposal in new mantis item. *** item 13 412 5.14.1 Q operator examples use aggregate constructors incorrectly Steve - the examples with {} are not quite right Dave - q concat in port expr or cont assign not that useful Gord - q of q allowed? how distinguish between adding a q q of integer q - use {q[1],q[2],q[3]} - what do with this? concat versus operation of adding a q to a q. - need to know type of LHS to do anything. - easiest work-around is to use methods - heavily overloading {} to express things naturally. The rules of what is going on here is not obvious. If we add in parameters and other things it would not be obvious what is going on. The assignment patterns were done to make things clear. Steve - do you flatten or not? are you adding one item? - there is no definition of this construct except for the examples. It isn't a concat nor an assignment pattern. - a cast is a conversion of a result, it doesn't tell you what it originally was. k) the ballot issues (positive vote) 13 412 5.14.1 Queue operator examples use aggregate constructors incorrectly 23 517 concatenation syntax usage instead of pattern assignment syntax 24 520 examples of queues assignments are not legal array assignments 94 511 type assignments) 96 516 5.7 and 5.8 description of type compatible arrays 97 518 queues, arrays, page 51) 98 519 empty array 99 521 queues, pattern assignment) 100 522 queues concatenation) 36 270 (enhancement -- randomize) ---------------------------------------------------- 122 251 (enhancement -- multiple user defined bins for cross 123 253 ( a variable has a coverpoint but is used directly in a cross ) 171 564 (cross program references) 187 594 (16.8 special syntax for accessing interfaces through clocking block 190 597 (unnamed clocking blocks in bnf) 200 608 (16.14.2 clocking block value resolution 201 609 (16.14 confusing meaning of ##cycle with clocking drive and two more items from Mantis 340 Q methods prototypes http://www.eda.org/svdb/bug_view_page.php?bug_id=0000340 339 Q methods typos - , versus : http://www.eda.org/svdb/bug_view_page.php?bug_id=0000339 4) Action items to close all acknowledged Mantis (SVDB) for p1800 5) Next meetings schedule (possible) Monday May 23rd, 2005 Monday June 6th, 2005 [DAC June 13--17 Anaheim] WILL BE SENDING EMAILS PER REQUEST FROM P1800. ============================================================== Action items: (May 9, 2005) 1) AI (Mantis 725): Arturo update proposals into combined/merged one for 725 2) AI (Mantis 714): Ray to modify the proposal to make corrections for a,b page 334, and the text under the example also needs to be deleted. Place the .pdf file there. 3) AI (Mantis 703): Ray to update the proposal with the friendly ammendment and place pdf file. 4) AI (item 95- Mantis 512): Brad - place the proposal to the mantis database. 5) AI (Mantis 236): Gordon add a bugnote, to effect that this was reviewed and discussed 6) AI (Mantis 730): Brad place the proposal in new mantis item. 7) AI: ALL REVIEW ALL APPROVED/CLOSED MANTIS items for editorial correction