GED2HTML: What's New in the Current Release

It's been quite a while since I released a new version. Mostly this has been because I have felt that the program has reached a terminal stage of development, where further incremental changes would tend not to improve the program. The current release was prompted by the fact that I have moved my distribution site and E-mail addresses to a new location, which required that I update the program and documentation accordingly. At the same time, I have incorporated a few minor bug fixes and other changes that I had made to my development source over the past few years. These are:

• There is now only a single Windows version, which should run on any recent version of Windows.
• A bug was fixed that resulted in a failure to link texts into the substructures of a source citation.
• Some changes to the default output program were made to address certain issues arising from malformed dates in the GEDCOM.
• A bug was fixed in the default output program that would cause single-letter surnames to appear in the wrong index tab group in the output.
• Some error messages were corrected.

This release consists of incremental updates and improvements I have been making since the release of version 3.5 a year and a half ago. The following changes were made since version 3.5e:

• I've decided to discontinue support for the Windows 3.1 version. I apologize to those who are still using Windows 3.1, but, hey, even Microsoft doesn't support it anymore.

• The documentation is now accessible from the program via a "HELP" button, using the standard Microsoft "HTML Help" viewer.

• New option variables INCLUDE_PATH, INCLUDE_IMAGE_EXT, and INCLUDE_INFO_EXT were added. The INCLUDE_PATH variable is a string that is automatically prepended by the standard output program to the GEDCOM identifier (RIN) for an individual, to construct the name of an associated external file to be included in the HTML output. The default value is the empty string. The INCLUDE_IMAGE_EXT and INCLUDE_INFO_EXT variables are strings that are automatically appended by the standard output program to the RIN. The default values are .img and .inc, respectively. The INCLUDE_IMAGE_EXT variable is for the first include, which is intended to be used to include an image of an individual. The INCLUDE_INFO_EXT variable is for the second include, which is intended to be used to include additional information about an individual.

• A new .exists qualifier was added for string values. If x denotes a string, then x.exists evaluates to non-null if the string is the name of an existing file, otherwise to null. The behavior of include was changed so that an error message is produced if an attempt is made to include a nonexistent file. The standard output program was modified to test whether a file exists before attempting to include it.

• New option variables IGNORE_NOTES and IGNORE_SOURCES were added. If set to non-null, then the program will completely ignore any notes or sources in the input GEDCOM, other than !LIVING notes. This behavior is different from the previously existing GENERATE_NOTES and GENERATE_SOURCES variables, because the latter variables only suppress the production of separate notes and sources files, but left the links intact in the other pages.

• Processing was added for the fields NPFX, GIVN, NICK, SPFX, SURN, and NSFX of the NAME structure. These fields were previously ignored.

• Processing was added for the fields ADR1, ADR2, CITY, STAE, POST, and CTRY of the ADDR structure. These fields were previously ignored.

• An option was added to put dates on the pedigree chart. The production of dates is controlled by the new option variable PEDIGREE_DATES, which has the default value 1, meaning that dates are produced.
• New qualifiers .christening and .burial were added for individual values. These produce events similarly to the previously existing .birth and .death qualifiers.

• A new statement, deftag, was added to the output language. The syntax is "deftag <id> <expr>, where <id> is a GEDCOM 5.5 tag (such as INDI or BIRT) and <expr> is an expression that evaluates to a string. The effect is that the default print name for the specified GEDCOM tag (such as "Individual" or "Birth") is replaced by the new string. This construct can be used by the end user to customize the GEDCOM tag names that get printed out with events; for example, if the user does not agree with the particular default translation of the tag name in a particular language.

• A new option variable NO_LDS_ORDINANCES was added. If nonzero, then the production of output corresponding to LDS ordinance information is suppressed. The default value is zero.

• Support was added for producing Slovenian output.

• Support was added for producing Portuguese output.

• Support was added for producing Afrikaans output.

• The method of sorting individuals in the index was modified so that individuals with identical names are sorted chronologically according to any available date information.

• The default output program was changed so that pedigree charts will not be printed for an individual if there is no information about that individual's mother or father.
• A problem was corrected in which the output interpreter failed to interpret array subscripts correctly.

• A bug was corrected that was causing the date in the "Created by" page footer to come out in the US English locale instead of any specified locale.

• The old locale names used under FreeBSD were updated to those currently in use in FreeBSD 3.x.

• A problem was corrected with the interpretation of GEDCOM dates that used the FROM and TO keywords.

• A bug was fixed that caused extra junk to be left at the end of a text field when @ was used in the GEDCOM.

• The method for converting strings to upper case was modified to have the correct results for accented letters. This conversion should now work correctly, at least for languages that use the ISO-8859-1 character set. Slovenian is the only supported language so far that does not use this character set.

• Various minor corrections and improvements were made in the standard output program.

Another mostly bugfix release that supersedes version 3.5d. The following changes were made since version 3.5d:

• Incorrect codes in internal tables for the ANSEL character set were corrected. This problem caused certain international characters, including "es zet", slash o, thorn, ligature ae, and various other symbols, to appear incorrectly in the output for certain GEDCOMs.

• Some minor changes were made to the Swedish language tables.

• Failure of GED2HTML to correctly interpret dates with trailing spaces was corrected.

• A missing comma was added to the output program. This was causing a mysterious "00" to be output after the LDS "Temple" tag.

• Father, mother, and child links from family group sheets were changed to point to places that were a bit more reasonable. In particular, an attempt is made to always point such links to the most suitable family group page, if such exists, giving preference to a page that records the marriage of the target individual over one that records that individual as a child.

• Code was added to the output program to link to notes and sources for "Married:" events, the same was as was already done for "Born:" and "Died:" events.

• Finnish was added to the list of supported languages.

• Some little-used and/or old GEDCOM tag names that appeared to be coded incorrectly in internal tables were corrected.

Another bugfix release that supersedes version 3.5c. The following problems in version 3.5c were fixed:

• A problem was corrected that caused the alphabetical index tabs not to point correctly to the proper surnames in case there were null surnames in the GEDCOM input.

• A problem was fixed in which children's marriage date/place information was coming out blank in the parents' family group sheet.

• In family group sheets, father/mother links now point to the page for the correct marriage, rather than to the first marriage for the specified individual.

• Superfluous blanks that were being output in the gendex.txt file were removed.

• A bug in the output language that prevented subscripted variables from working properly was corrected.

Another bugfix release that supersedes version 3.5b. The following problems in version 3.5b were fixed:

• The "standard.g2h" output program was reworked to fix brokenness in the support of the INLINE_NOTES option, and to make the INLINE_NOTES and INLINE_SOURCES options behave like the documentation says they are supposed to.

• A problem was fixed in the "standard.g2h" output program that caused it to fail during generation of the alphabetical surnames index in cases where there were individuals with no surnames.

• The error message claiming that a date is "missing non-GEDCOM-5.5-compliant: was toned down to state only that the the date is "missing or uninterpretable". This is because there are a couple of constructions in GEDCOM-5.5 dates that GED2HTML does not yet understand, so it seems best not to declare arrogantly that they are not GEDCOM-5.5-compliant.

• The behavior of the GED2HTML.EXE dialog screen was changed so that selecting an options file through the use of the BROWSE button causes that options file to be loaded immediately. The failure of this loading to occur has confused many people in the past.

This additional bugfix release supersedes version 3.5a. The following problems in version 3.5a were fixed:

• A problem was fixed in which the day of the month portion of dates failed to appear in the HTML output.

• A problem was fixed that caused error messages and failure of the program to generate output when the SURNAME_WIDTH parameter was set to a nonzero value.

• A problem was corrected that caused children in family group sheet output to be listed with the same marriage place as their parents.

• The number of error messages printed out in case of non-GEDCOM-5.5-compliant date strings was limited, with full details going to the error file.

This bugfix release supersedes version 3.5. The following problems in version 3.5 were fixed:

• A problem was fixed in which notes that had CONC in them were being truncated after two lines.

• A workaround was implemented for the situation in which the GEDCOM file contains non-GEDCOM-5.5-compliant date formats. (Brother's Keeper appears to be a significant culprit here.) GED2HTML v3.5 would silently interpret such dates incorrectly. Version 3.5a now complains loudly about the incorrect date format, and then simply uses the date string verbatim, rather than trying to interpret it. This means that non-compliant dates will not be translated according to the selected output language or locale.

• A mistake was corrected regarding the transmission of the "Output Program File" parameter between the GED2HTML.EXE dialog screen and the G2H.EXE program. The dialog screen was using the old parameter name TEMPLATE_FILE, rather than the new parameter name OUTPUT_PROGRAM.

It's been a long time since version 3.1a, and as I've made a fairly substantial number of improvements, I've bumped the version number several notches. I didn't go all the way to 4.0, though, because the changes to the program are primarily evolutionary, rather than revolutionary. The following summarizes the most important changes. There have also been many other minor changes and fixes.

• The family group sheet output format has been integrated into the standard output template program, and it is selectable via the "Additional Options" field of the dialog box.

• A number of parts of the GEDCOM 5.5 specification that were not handled before, or were handled only minimally, are now handled more fully. This includes sources and source citations, as well as repository information.

• Dates in the GEDCOM are now interpreted by GED2HTML, if possible. This enables dates to be produced in the HTML output in a format appropriate for the desired output language (English, French, etc.)

• The separate output programs for each individual output language have been merged into a single standard program. This makes it possible to change the output language via program options, without having to also use a different output program. The disadvantage is that the standard output program is now a daunting piece of code of over 2500 lines, which probably makes it fairly difficult for non-programmers to make substantial modifications.

• The output template now provides some control, via options variables, of whether notes and sources appear on the same page as the individual or family to which they pertain, or else on a separate notes page.

• For surname indexes that aren't too long, the standard template now uses an alphabetical style that seems to be popular on the Web. To support this, the output template language was extended with comparison operators on both integer and string values.

• Some non-backward-compatible changes were made to the output language in order to support some of the new features in this release. These changes primarily have to do with the additional of additional types of values, and changes in the meanings of certain qualifiers. If you have made substantial customizations to the output program, you will want to review the output language definition.

It's sort of embarrassing to issue a new release a few days after a previous release, but in this case I had to do it because of a bug in the new "automatic living detection" feature that could potentially cause me to get a lot of E-mail.

• A bug was fixed that prevented many dates from being interpreted correctly, which would in turn cause individuals that should have been presumed "living" not to be flagged properly.

This release consists of minor bug fixes and changes I introduced in response to E-mail from users, plus a new experimental output format, ability to process LDS ordinance information, and a feature for automatic flagging of "living" individuals.

• A bug was fixed where when an individual had multiple NAME records, the last one was chosen as the primary name, rather than the first.

• The standard template was modified to add an indication of whether the family line continues with a child.

• A new experimental template "famgrp.g2h" was added, which produces the individual data pages in "family group" format, much in the style of another popular GEDCOM to HTML converter program. Although the fact that I could code this template illustrates the flexibility of the GED2HTML output processor, it wasn't as straightforward as I would have liked. In the future, I will be looking into generalizing the output language to make it easier to generate output formats radically different than the "original" GED2HTML format.

• Some problems with umlauts in ANSEL-encoded GEDCOMs were fixed.

• Processing was added for LDS ordinance information.

• A bug was fixed that caused a GPF or invalid page fault when processing a GEDCOM that had undefined cross-reference links to individuals.

• Several new options variables were added that make it easier to set background colors or images on the generated HTML pages.

• Added a feature by which GED2HTML will automatically presume a person to be living if they are born after a specified year and have no explicit death information.

• Support for output in German and Norwegian has been added.

• The program is now somewhat more forgiving in case there are undefined cross-reference ID's in the GEDCOM. Previously, it would often terminate prematurely in the output phase. Now, it should complete producing the output files in most situations.

• A bug was fixed that prevented options beginning with a digit from being recognized as strings, even when there was a preceding double quote.

• Some improvements and extensions to the standard template program.

This version incorporates a few bug fixes and extensions since version 3.0. Specifically, the following changes were made:

• The processing of "individual attributes", such as EDUC, RELI, RESI, was broken. These should now work much better.

• The GED2HTML dialog box would mysteriously fail to save options settings in certain situations. This was due to a parameter being set too low, which limited the number of changed options that could be recorded in an options file to 10. I have now increased this parameter to 100.

• I got rid of a long directory name (templates) that was causing certain unzip programs to choke on the distribution file. The template directory is now called "template".

Version 3.0 represents a major rewrite of a large portion of GED2HTML. The main objective of this rewrite was to modify the internal design of the program to obtain flexibility for accomodating the GEDCOM 5.5 standard as fully as possible. There were three main problems with the old design that stood in the way of doing this: (1) the output template language was too complicated, making it infeasible to add further features or information to the output pages; (2) the internal data structures could not accomodate the additional information allowed by GEDCOM 5.5 without using excessive amounts of memory; (3) the method of specifying program options was too complicated and ad hoc to support the introduction of additional options for output customization. The following is a brief summary of the main changes in functionality between 2.5b and 3.0:

• The output template language has been completely redesigned. The new language is much more like an ordinary programming language, and it actually has some features for finding errors in templates before they are run, as well as for debugging templates while they are running. The several output templates that existed in the previous version have been combined into a single template, and recoded in the new language.

• The method by which customization options are set has been completely revised and made more uniform. Nearly all options are now controlled by setting variables in the output template interpreter. Many more customization options were added.

• Under Windows, the GED2HTML.EXE dialog box has been redesigned and substantially extended, so that essentially any program option can be set from it. It is now possible to save options settings to a file and recall them on a later run, making it much easier to associate a different collection of options settings with each GEDCOM one might wish to run. The dialog box has been enhanced with "Browse" and "Save As" common dialog boxes for specifying the GEDCOM, options, and template files.

• The program now has a native Windows 95 version, as well as a Windows 3.1 version. The Windows 95 version is several times faster than the Windows 3.1 version.

• An attempt has been made to make GED2HTML compatible with the current GEDCOM 5.5 standard. Besides INDI and FAM, both both NOTE and SOUR at level zero are now fully supported. Though other level-zero records permitted by GEDCOM 5.5 are not yet supported, the internal design of the program has been reworked so that it should be much easier to add support for these records in future releases.

• Support for several character sets other than ISO-8859-1 has been added. In particular, the program should now do a reasonable job with international characters encoded in ANSEL, ISO-8859-1, and IBMPC character sets under DOS code pages 850, 863, and 437.

• The program now supports output in French and Dutch as well as in English.

• A two-level hierarchical surnames index is now supported, to reduce the size of the surnames index files for very large databases.

• Notes and sources files can be split up into pieces, so that each piece is a manageable size.

• The registered and "demo" versions are now identical, except for the following: (1) the demo version outputs an UNREGISTERED message on each HTML output page; (2) substitution of new output template programs is not supported in the demo version; (3) the demo version will read in, but not produce output for, GEDCOMs containing over 5000 individuals.

• It is now possible, using an option setting, to switch between the hash-based assignment of individuals to files and directories of version 2.5b, and the alphabetical assignment used in older versions. This should alleviate a common complaint about 2.5b.

• The behavior of the !LIVING feature has been placed under the control of the output template program. The default behavior is to output the name of a living individual, but no other information. More or less information can be included by changing the output template program.

• Better error messages are produced in response to problems with the GEDCOM being processed.

• It is now possible to access some of the information in the 0 HEAD GEDCOM header record.

• Fixed bug (again!) which caused earlier Windows versions to look for !INCLUDE files in either the directory below (versions earlier than 2.5) or above (versions 2.5 and 2.5a) the directory where the documentation says it looks.
• Fixed a bug which caused surname ``JAMES'' to appear before surname ``JAME'' in the index.
• Fixed a bug in the that was causing links between sources to fail.
• Fixed an omission from the output templates that was causing sources associated with family events not to appear in the HTML output.
• Added some documentation about international character sets, and revised some other parts of the documentation.

• Fixed -w 0 option so it works instead of crashing the program.
• Added a trailer line to the GENDEX files to help in detecting truncation.
• Added Kagi Shareware REGISTER.EXE program to provide additional ways to pay the registration fee under Windows. This should make it feasible and convenient for non-US users to pay by credit card.

• Revamped the scheme for assigning individuals to subdirectories and files within subdirectories. The new scheme calculates the particular directory and file in which to put an individual from the individual's cross-reference ID, so the same individual will always end up in the same directory and file as long as the same number of directories and number of files per directory are used. A new option (-D n) was added to allow the user to specify the number of directories to be used. This should make it possible to link to individual pages from external documents, and have these links remain valid when the GEDCOM is re-processed, even if individuals have been added to the GEDCOM since the first run.
• Added the following feature to aid in eliding information about living persons: If an 0 INDI record for an individual contains a field that begins 1 NOTE !LIVING, then all personal data is deleted from that individual's page and the only information that is output is the text of the note.
• Eliminated large array allocation that was effectively limiting the number of individuals that could be processed to 16384 under Windows.
• Changed the option syntax for dumping compiled-in templates so that it can actually be used under Windows to dump templates to a file.
• Fixed a problem with empty surnames generating multiple entries in the surnames index.
• Fixed a problem under Windows, where files to be included in the HTML output were being looked for in the wrong directory.
• The GENDEX.txt index files now use christening date/place if no birth information is available, and burial date/place if no death information is available.

• Fixed bug that could cause infinite looping at end of file on some Unix systems, and caused weird syntax errors when 8-bit characters occurred in GEDCOM files under Windows.
• Fixed bug where the alternate index and surname template fields were interchanged in the GED2HTML.EXE dialog box under Windows.
• Windows executables are much smaller, and probably run a bit faster, due to the use of different compiler options.

• Introduced changes that cut memory usage in half, and also significantly boost performance.

• Some progress printout was added to the Windows version, because people were mailing me telling me that the program had hung up, when in fact it was just taking a long time.

• GED2HTML is now being released as shareware. Please see here for more information.
• New documentation has been written in HTML. You're reading it.
• The DOS version of the program has been discontinued. Only the Windows version is in the current release. The DOS version was pretty useless for GEDCOM's of any interesting size, anyway.
• A true Windows user interface is now provided to make it easier and less confusing for most Windows users to run the program.
• Source records are now supported. Previous versions of the program would only include source information at level 1 within INDI records. Furthermore, full source structures were not supported. The Version 2.4 handles cross-referenced source records at level 0, as well as source structures within INDI, FAM, NOTE, and event structures at higher levels. The existence of source information is signalled in the output file by ``footnote markers'' that appear near the information to which they pertain. The footnote markers link to source information in separate HTML files. The chaining of source structures, as described in the GEDCOM 5.3 draft standard, is also supported, but it is not thoroughly tested since I do not have sample GEDCOMs that use this feature. At present, only source structures chained to other source records are handled. Chaining of source structures to event records, though allowed by the GEDCOM 5.3 draft standard, will result in an error message.
• Performance improvements have been made in the production of the output files, especially under Windows.
• OCCU, EDUC, and a variety of other tags giving information about an individual are now processed properly, and the information appears in the output files.
• Pedigree charts have been changed slightly to be a little more compact at depths of 4 or greater.
• PERSONS.html is now the name of the top-level index file, rather than INDEX.html. The latter was an unfortunate choice that conflicted with defaults built into many Web servers.
• Titles of individual pages now include birth and death dates.
• New template constructs have been added to cause the date of the run and the version number of GED2HTML to be inserted into the output.
• The Windows version of the program now creates its output files in a subdirectory HTML, rather than in the directory from which it was run.
• The hierarchical index algorithm was modified to try to reduce the frequency of index nodes with hardly any entries in them. The current scheme adjusts the width of the index nodes less than the specified maximum width, so that all the nodes at the same level of the index have close to the same number of entries. Under certain conditions the top-level index node can still end up with as few as two entries, but this can be worked around by adjusting the specified maximum width until better results are obtained.
• A few minor bugs were fixed that existed in the previous version of the program.