OASIS Open Document Format for Office Applications (OpenDocument) TC

Hi all, what do you think about the proposal below? I'm not sure about child elements <text:h>. LibreOffice does not create such child element in a spreadsheet. When reading a file or copy&paste a table from a text document, LibreOffice treats an <text:h> element same as an <text:p> element and saves it as <text:p> element. Kind regards Regin Proposal ======== 4.7 empty cell The technical term "empty cell" is used in this part of the specification in defining functions and expressions. The technical term "empty cell" stands for a cell, for which non of the following properties is true: The cell has a table:formula attribute. The cell has a <text:p> child element. The cell has an office:value-type attribute and the corresponding value attribute office:value, office:date-value, office:time-value, office:boolean-value, or office:string-value, respectively. Note: The markup <table:table-cell><text:p/></table:table-cell> generates no "empty cell" but a cell containing an empty string.

Hi, For me (and Gnumeric), this is an empty cell: <table:table-cell  office:value-type = string  table:style-name = ce2 > <text:p/> </table:table-cell> After reading <table:table-cell  office:value-type = string  table:style-name = ce2 > <text:p/> </table:table-cell>  and <table:table-cell  office:value-type = string  table:style-name = ce2 > </table:table-cell> are indistinguishable. The <text:p/> element exists only as markup in the saved file, not in the idealized spreadsheet. Wanting to distinguish between these two requires that otherwise completely identical cells would have meta flag whether they are empty or not. Andreas On 2020-05-03 2:21 p.m., Regina Henschel wrote: Hi all, what do you think about the proposal below? I'm not sure about child elements <text:h>. LibreOffice does not create such child element in a spreadsheet. When reading a file or copy&paste a table from a text document, LibreOffice treats an <text:h> element same as an <text:p> element and saves it as <text:p> element. Kind regards Regin Proposal ======== 4.7 empty cell The technical term empty cell is used in this part of the specification in defining functions and expressions. The technical term empty cell stands for a cell, for which non of the following properties is true:  The cell has a table:formula attribute.  The cell has a <text:p> child element.  The cell has an office:value-type attribute and the corresponding value attribute office:value, office:date-value, office:time-value, office:boolean-value, or office:string-value, respectively. Note: The markup <table:table-cell><text:p/></table:table-cell> generates no empty cell but a cell containing an empty string. --------------------------------------------------------------------- To unsubscribe from this mail list, you must leave the OASIS TC that generates this mail. Follow this link to all your TCs in OASIS at: https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php -- Andreas J. Guelzow, PhD FTICA Registrar & Director of Enrolment Services Professor, Mathematical & Computing Sciences Direct: +1 780 479 9290 Toll-Free: +1 866 479 5200 concordia.ab.ca

Hi Andreas, Andreas J Guelzow schrieb am 03-May-20 um 23:51: Hi, For me (and Gnumeric), this is an empty cell: <table:table-celloffice:value-type="string"table:style-name="ce2"> <text:p/> </table:table-cell> After reading <table:table-celloffice:value-type="string"table:style-name="ce2"> <text:p/> </table:table-cell> and <table:table-celloffice:value-type="string"table:style-name="ce2"> </table:table-cell> are indistinguishable. The <text:p/> element exists only as markup in the saved file, not in the idealized spreadsheet. Wanting to distinguish between these two requires that otherwise completely identical cells would have meta flag whether they are empty or not. Was it changed in Gnumeric? I have tested it with the GnumericPortable_1.12.17 for Windows (newer one does not exist). For a cell A2 generated from markup <table:table-cell table:style-name="ce2" office:value-type="string"> <text:p/> </table:table-cell> the function ISBLANK(A2) returns FALSE here. Kind regards Regina

HI, On 2020-05-03 6:26 p.m., Regina Henschel wrote: Hi Andreas, Andreas J Guelzow schrieb am 03-May-20 um 23:51: Hi, For me (and Gnumeric), this is an empty cell: <table:table-celloffice:value-type= string table:style-name= ce2 > <text:p/> </table:table-cell> After reading <table:table-celloffice:value-type= string table:style-name= ce2 > <text:p/> </table:table-cell> and <table:table-celloffice:value-type= string table:style-name= ce2 > </table:table-cell> are indistinguishable. The <text:p/> element exists only as markup in the saved file, not in the idealized spreadsheet. Wanting to distinguish between these two requires that otherwise completely identical cells would have meta flag whether they are empty or not. Was it changed in Gnumeric? I have tested it with the GnumericPortable_1.12.17 for Windows (newer one does not exist). For a cell A2 generated from markup  <table:table-cell table:style-name= ce2 office:value-type= string >  <text:p/>  </table:table-cell> the function ISBLANK(A2) returns FALSE here. I do not recall changing that. In current Gnumeric (1.12.47) it returns TRUE. Gnumeric is only released in source and it is not supported on Windows. 1.12.17 is from June 2014. Most of the import of cell content text formatting was written more recently than 1.12.17. On import from ODF GNunmeirc converts everything into its own data structures and it is not clear to me how, in 1.12.17,  <table:table-cell table:style-name= ce2 office:value-type= string >  <text:p/>  </table:table-cell> would be distinguished from  <table:table-cell table:style-name= ce2 office:value-type= string >  </table:table-cell> after import. Andreas Kind regards Regina --------------------------------------------------------------------- To unsubscribe from this mail list, you must leave the OASIS TC that generates this mail. Follow this link to all your TCs in OASIS at: https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php -- Andreas J. Guelzow, PhD FTICA Registrar & Director of Enrolment Services Professor, Mathematical & Computing Sciences Direct: +1 780 479 9290 Toll-Free: +1 866 479 5200 concordia.ab.ca

The attached file has in A1: <table:table-cell table:style-name= Gnumeric-default ><text:p/></table:table-cell> and in A2: <table:table-cell table:style-name= Gnumeric-default /> B1 and B2 have formula calls to determine whether A1 and A2 respectively are blank. When I open the file in Gnumeric 1.12.47 they both return TRUE. When I open the file in Libreoffice 5.1.6.2 I also get TRUE !? Andreas On 2020-05-03 6:26 p.m., Regina Henschel wrote: Hi Andreas, Andreas J Guelzow schrieb am 03-May-20 um 23:51: Hi, For me (and Gnumeric), this is an empty cell: <table:table-celloffice:value-type= string table:style-name= ce2 > <text:p/> </table:table-cell> After reading <table:table-celloffice:value-type= string table:style-name= ce2 > <text:p/> </table:table-cell> and <table:table-celloffice:value-type= string table:style-name= ce2 > </table:table-cell> are indistinguishable. The <text:p/> element exists only as markup in the saved file, not in the idealized spreadsheet. Wanting to distinguish between these two requires that otherwise completely identical cells would have meta flag whether they are empty or not. Was it changed in Gnumeric? I have tested it with the GnumericPortable_1.12.17 for Windows (newer one does not exist). For a cell A2 generated from markup  <table:table-cell table:style-name= ce2 office:value-type= string >  <text:p/>  </table:table-cell> the function ISBLANK(A2) returns FALSE here. Kind regards Regina --------------------------------------------------------------------- To unsubscribe from this mail list, you must leave the OASIS TC that generates this mail. Follow this link to all your TCs in OASIS at: https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php -- Andreas J. Guelzow, PhD FTICA Registrar & Director of Enrolment Services Professor, Mathematical & Computing Sciences Direct: +1 780 479 9290 Toll-Free: +1 866 479 5200 concordia.ab.ca Attachment: Empty-2choices.ods Description: application/vnd.oasis.opendocument.spreadsheet

Hello Andreas, in this case, I assume you haveÂfound two anomalies. :-) Allow me to explain the problem from a different angle: In the past years of my absence to the ODF TC, I have been working on the standardization of EU invoices. CEN, the European standardization norm, has specified for the business to government trade within Europe an obligatory semantic model that can be serialized as two different XML formats (UBL & UN/CEFACT XML). The serialization, in theory, does not matter as long it can be normalized and transformed back to the unique obligatory semantic model. Back to ODF. Let me ask a similar question as Regina did: "What is an empty document?" An easy and correct answer is that smallest amount of ODF XML that makes a valid ODF document of a certain type according to our ODF XML RelaxNG grammar. But are there multiple answers possible? If they can be normalized to one XML instance, that has not a different semantic. Some ODF applications insert an empty paragraph by default into the empty text document. They can do this, but it is not in fact equal to an empty document,Âeven if some ODF applications behave similarlyÂto the two documents. Please keep in mind that I am still working on change-tracking and that the addition of an empty paragraph does make a difference to the document state. I have created a test implementation that transforms an ODT document into an equivalent list of ODT changes (serialized in JSON, here is every line a user change) . The existence or none existence of the empty paragraph within the empty document would alter the list of changes, therefore, the user semantic. Does this help you to answer the earlier question? Regards, Svante Am Mo., 4. Mai 2020 um 03:48ÂUhr schrieb Andreas J Guelzow < andreas.guelzow@concordia.ab.ca >: The attached file has in A1: <table:table-cell table:style-name="Gnumeric-default"><text:p/></table:table-cell> and in A2: <table:table-cell table:style-name="Gnumeric-default"/> B1 and B2 have formula calls to determine whether A1 and A2 respectively are blank. When I open the file in Gnumeric 1.12.47 they both return TRUE. When I open the file in Libreoffice 5.1.6.2 I also get TRUE !? Andreas On 2020-05-03 6:26 p.m., Regina Henschel wrote: Hi Andreas, Andreas J Guelzow schrieb am 03-May-20 um 23:51: Hi, For me (and Gnumeric), this is an empty cell: <table:table-celloffice:value-type="string"table:style-name="ce2"> <text:p/> </table:table-cell> After reading <table:table-celloffice:value-type="string"table:style-name="ce2"> <text:p/> </table:table-cell> and <table:table-celloffice:value-type="string"table:style-name="ce2"> </table:table-cell> are indistinguishable. The <text:p/> element exists only as markup in the saved file, not in the idealized spreadsheet. Wanting to distinguish between these two requires that otherwise completely identical cells would have meta flag whether they are empty or not. Was it changed in Gnumeric? I have tested it with the GnumericPortable_1.12.17 for Windows (newer one does not exist). For a cell A2 generated from markup  <table:table-cell table:style-name="ce2" office:value-type="string">  <text:p/>  </table:table-cell> the function ISBLANK(A2) returns FALSE here. Kind regards Regina --------------------------------------------------------------------- To unsubscribe from this mail list, you must leave the OASIS TC that generates this mail. Follow this link to all your TCs in OASIS at: https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php -- Andreas J. Guelzow, PhD FTICA Registrar & Director of Enrolment Services Professor, Mathematical & Computing Sciences Direct: +1 780 479 9290 Toll-Free: +1 866 479 5200 concordia.ab.ca --------------------------------------------------------------------- To unsubscribe from this mail list, you must leave the OASIS TC that generates this mail. Follow this link to all your TCs in OASIS at: https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php

Regina Henschel schrieb am 03-May-20 um 22:21: Hi all, considering the discussion, here the next proposal. Still unclear how to handle <text:h> child element. Excel can produce a cell with empty string by entering an apostrophe. While you are working on that spreadsheet, ISBLANK() returns FALSE for that cell. It is written out as markup <table:table-cell office:value-type="string" table:style-name="ce2"><text:p/></table:table-cell>. After reopening the file, ISBLANK() returns TRUE. So I consider this an error in Excel import. Kind regards Regina New Proposal ============ 4.7 empty cell The technical term "empty cell" is used in this part of the specification in defining functions and expressions. The technical term "empty cell" stands for a cell, for which non of the following properties is true: The cell has a table:formula attribute. The cell has a non-empty <text:p> child element. The cell has an office:value-type attribute and the corresponding value attribute office:value, office:date-value, office:time-value, office:boolean-value, or office:string-value, respectively. The cell has an office:value-type="string" attribute and a <text:p> child element. In case of an empty child element <text:p/>, the cell contains an empty string. It is implementation-dependent whether the markup <table:table-cell><text:p/></table:table-cell> is an "empty cell" or not. For interoperability producers should not write such markup, but add an office:value-type="string" attribute in case a cell with empty string is intended, or omit the empty child element <text:p/> in case an "empty cell" is intended.