Known Outstanding
Problems
Quotation Marks in Original Titles
The current approach for listing previous titles inside double quotation marks says that if the original title contained double quotation marks then they should be changed to single quotation marks. However, there is no way of handling titles containing single quotation marks (e.g. "The `Z' Man" by Leslie Charteris). (Requires data changes)
The date formats used on "en" and "pu" records are non-standard and would be better replaced with dates in the usual format. (Would require data changes - the current intention is to eliminate these as far as possible)
Cartoonists are currently listed as "authors" rather than "artists". Some thought needs to be given to this. (Fixed in v2)
CFI: Edward Holmes' serial "The Trail of the Missing Million" ran in The Thriller Library in 5 parts, the first three of which were reprints and the final two original. However in the Author Index listing there is no mention of the reprint appearance of the first three parts. (Fixed in v2)
CFI: The entry in the Author Index for "No. "000"", and a few others, has a semi-colon between the name and the dates for no apparent reason. (Fixed in v2)
If a story published under a pseudonym is reprinted under the author's real name, the pseudonym entry in the Author Index contains the "final version" of the real author's name, while the cross-references use the "raw version" of the name, which looks odd (e.g. "Ramon Decolta" on S215.HTM in the CFI is a pseudonym of "Raoul F. Whitfield" but has references to "Raoul Whitfield").
If a story by a particular author was originally published as a collaboration with another author, the Author Index ends up with an entry under the former author saying you should look under the current entry (e.g. "The Riddle of the Silver Death" under the Thomas W. Hanshew entry in the Author Index in the CFI says "see also under Thomas W. Hanshew"). (Fixed in v2)
When a concatenated entry (such as a column) appeared in a given magazine with different co-authors it would be preferable if all occurrences with a particular co-author were listed together rather than potentially being split into multiple sections based on the chronological appearance (e.g. "Oddities in Crime" by Ejler Jakobsson in the Author Index in the CFI). (Fixed in v2)
When a concatenated entry ends with one or more unpublished issues the " (unpublished)" is attached to the year rather than to the issue(s) concerned (e.g. "Murder-Go-Round" by Harriet Klausner in the Author Index in the CFI). (Fixed in v2)
When the real author behind a house name is doubtful, the author hyperlink for the real author omits the trailing ", [?]" which looks rather odd in context (cf. C.K.M. Scanlon's "Big Shot" in the Author Index in the CFI). (Fixed in v2)
If a story is published under a house pseudonym, then it is listed in the Author Index under both the pseudonym and the real name; if it is published under a personal pseudonym then it is listed only under the pseudonym. Thus, for example, the entry for Sam Merwin, Jr, lists the dozen Mike Shayne stories he wrote as Brett Halliday, while the entry for Davis Dresser only lists a handful of the Mike Shayne stories he wrote that were published under his own name. It would be desirable if the two were treated the same way. (Fixed in v2)
When a story has multiple authors, the order of the authors is preserved correctly in the T*.HTM but appear to be reorganised into alphabetical order in the L*.HTM and F*.HTM files, which seems odd (cf. "From Watson's Scrapbook" and "Haven't Seen You Since the Funeral"). There needs to be a way to suppress this (with a diagnostic trap for variations).
In the CLN* pages, magazines that have no concrete data, or are phantom, or are cross-references to other titles, have no dates included which looks odd. It would be useful to have some way of including dates for these entries as well. Conversely, magazines which already have dates in the header records end up with duplicated dates which also looks odd.
When an item is extracted from a larger work (e.g. "Clarence Darrow" in EQMM 10/83) it would look better if the "from <xxx>" were moved to follow the publication data in the same way as author and title changes are now handled. (Fixed in v2)
While the author sorting strips out punctuation before sorting, the same does not happen for story title sorting. Ideally this should also ignore punctuation (see, for example, L1.HTM in the CFI where "A.B.C." is nowhere near "The ABC Code" and where "Abra Cadaver!" comes before rather than after "Abracadabra"). (Fixed in v2)
The "partial capitalization" of author names in the author index can lead to some very odd results and needs rethinking, ideally allowing the compiler to select from multiple algorithms. (Fixed in v2)
When concatenating, author notes like "as told to:" and "tr:" should be ignored (e.g. Chile Acuna "Memoirs of a Police Informer" in the Author Index in the CFI).
When an item is reprinted under a different name (e.g. "The House in Goblin Wood" by Carter Dickson, reprinted as by John Dickson Carr) (Fixed in v2)
It would be preferable to have a compiler-configurable control file to define index title, sub-indexes required, optional behaviour, etc. etc. rather than hard-wiring these into the programs. It would also be desirable to allow additional flexibility within individual indexes, such as:
The programs need more "defensive" coding that validates the data as it processes it, flagging up any anomalies, rather than just assuming the data follows a particular format (e.g. unexpected author note types can lead to currupted entries, unbalanced braces or brackets can lead to vast swathes of the index appear in bold and/or italics). (Fixed in v2)
Currently the programs are case-sensitive in some parts and case-insensitive in authors. This needs to be rationalised. (Fixed in v2)
There are currently problems if a magazine name changes partway through a serial or column. (e.g. E.R. Home-Fall & "The Mounties Kicked Him Out") (Fixed in v2)
The HTML generated by the programs uses a number of Windows-specific characters; it would be preferable to use a more portable mechanism (such as UTF-8) throughout. (Fixed in v2)
A number of series which have been written by mutliple authors have a clearly identified author who is the "key author" for the series (e.g. Arthur Conan Doyle and the Sherlock Holmes stories). It would be desirable for stories by this author to be listed before all other authors in "by author" series listing. (Fixed in v2)
The chronological indexes have problems handling magazine issues identified by issue or volume number. (Fixed in v2)
In the B*.HTM pages it would be better if the "--- see under xxx" references were on the same line as the main title (e.g. Detective Romances).
Items that are specified with a column or series heading (i.e. column|item or series||number|item) are listed in the index in two parts with the column/index heading first followed by the item name(s) indented by a leading "___". This works very well in the author and title indexex if a single author has multiple items with the same heading but works less well when the series is a multi-author one such as "Adventurers All" - see, for example, the entry for R. Caton Woodville in the Author index in the FMI. Conversely (I think) the sub-heading is never listed in the Issue Index even if there are multiple adjacent entries.
It would be preferable if there were a consistent rule that said that the sub-heading approach would be used if, and only if, there are 2 or more adjacent items with the same heading. This would also remove the need for the 'Z' control code in the series control file which was implemented purely to address this problem.
If a magazine has "First March" issues (and such-like) the dates are currently displayed as "Mar #1, 1932" or similar (e.g. "The Gold Spike" under Jared L. Fuller in the Author index in the FMI). This would be better as "1st March 1932" or similar. (Fixed in v2)
Currently, if a serial has more than 9 parts, an additional space needs to be added to the "[Part n of m]" so that the parts are sorted into the correct order. It would be better if the programs inserted this automatically as and when needed.
There is currently a conlict when
generating author name cross-references for similar names (e.g. "Peter
M. Adlard - see under Mark Adlard") when both versions of the name are
clear from the main record (i.e. in this case the entry reads "ADLARD,
MARK; [i.e., Peter Marcus Adlard]"). If the two entries would appear together
(or very close) then the "see under" is redundant, and looks odd,
as the user's eye will already be drawn to the correct entry; if the two entries
are separated by several other entries then a "see under" would be
desirable. It would be better if this were a configuration option on an index-by-index
basis indicating how many intervening entries are needed before a "see
under" link would be generated (such that 0 meant the links were always
generated and 99999 meant they were never generated.
Phil S-P (11-Sep-2013):
I think that one core problem is that, no matter how good a reason we have for doing it, an alphabetical index that reads:
- ANDREWS, DAVID
- ANDREWS, GERALD
- ANDREWS, (Charles) ROBERT D(ouglas Hardy)
- ANDREWS, TIMOTHY
just looks wrong as the eye is drawn more to the names following the "ANDREWS, " rather than specifically to the names in capital letters. In particular, it can easily fool the eye of somebody looking for "Andrews, Robert D.", requiring two-or-three ranging scans to find the entry.
<<No "obvious" solution so left unresolved>>
Bill C. (4-Nov-2015)
E 11A0~Heald, Hazel/Lovecraft, H. P.~Horror in the Burying-Ground [as by Hazel Heald]~ss1937WRTMay~The ~
results in:
11 · The Horror in the Burying-Ground · Hazel Heald & H. P. Lovecraft · ss Weird Tales May 1937, as by Hazel Heald
HEALD, HAZEL (1896-1961) (chron.)
* The Horror in the Burying-Ground, (ss) Weird Tales May 1937; see also under
Hazel Heald.
* The Horror in the Burying-Ground (with H. P. Lovecraft), (ss) Weird Tales
May 1937, as by Hazel Heald.
Harper’s New Monthly Magazine May 1850
LOVECRAFT, H(oward) P(hillips) (1890-1937);
(about) (chron.)
* The Horror in the Burying-Ground (with Hazel Heald), (ss) Weird Tales May
1937, as by Hazel Heald.
Harper’s New Monthly Magazine May 1850
There is a dual entry under Heald because when the author and/or title of a reprint are different from the original publication the programs generate an entry as it appeared in the original publication with the note "see also under ..." added (if just the author is different). This provides a link between the different versions of the story. For an example of how this more commonly works see John Dickson Carr's "The Crime in Nobody’s Room"(http://www.philsp.com/homeville/cfi/s152.htm#A3132).
This does show that there are two different versions of how the author of the story is listed, but it also could be considered redundant. To eliminate the first entry under Heald I could put in a check for the the case where the author and "see also under" name are the same, but there may be cases where we'd want to list it.
(Fixed in v2)
If an author publishes a story under a house name, then that story is listed in the author index under both the house pseudonym and under the original author's name, e.g.
In addition, if an author publishes a story under a personal pseudonym and that story is later reprinted under the author's own name, then the original appearance and pseudonym is also listed under the original author's name, e.g.
However, any stories written under a personal pseudonym and not reprinted under the author's real name, and any stories reprinted under a personal pseudonym, are not listed under the author's real name. This means there is no single place where all of an author's story appearances can be found and is contrary to "common practice" (e.g. the ISFDb). The same problem exists with authors whose name changed (e.g. due to marriage) during their career.
A particular side-effect of this relates to the handling of uncredited items for an author whose name changed during their career as there is no way of knowing which name should be specified in the "Anon. ,(by:xxx)" field and if one indexer specifies one name in one magazine issue and another indexer specifies a different name in a different issue, then the two printings will not be identified as the same by the program. A similar problem exists where an author is better known by a pseudonym (such as with George Eliot) as any anonymous printings which specified the author's real name (i.e. "Anon. ,(by:Evans, Mary Ann)") would not match up with subsequent printings under the George Eliot name.
The ideal solution to address these problems would be:
This would mean that whatever name a user looked under they would find a consistent listing of all relevant data, and would (hopefully) resolve all issues related to the use of different names on different printings.
(Partially fixed in v2 - first part only)
When book reviews were first added to the database as part of indexing some of the British magazines (specifically Interzone and Astounding (UK)) a simple two record approach was used whereby the reviewer and the author/title of the book being reviewed were specified on a normal item record and publication details of the book concerned were specified on a separate, formatted, publication details record. Thus a simple example might be:
E _50A0~Miller, P. Schuyler~Modern Science Fiction~br1953ASFJun~~~~~Bretnor, Reginald~
E _50Q1~Coward-McCann~1953~52-11714~$3.75~274~
Some years later, rights were acquired to add information from Hal Hall's Science Fiction Book Review Index to the database, and these were added in a completely different format where all the information was squeezed into the item record with no clear format rules. Thus examples might be:
E _71A0~Miller, P. Schuyler~Modern
Science Fiction>, ed. Reginald Bretnor; New York: Coward-McCann, 1953. 294
p. 52-11714.~br1953ASFJun~<~~~~Bretnor, Reginald~
E_182A0~Sakers, Don~Transpecial>, Jennifer R. Povey; CreateSpace, 218pp,
$11.99, tp, ISBN: 978-150864746-1~br2016ASFJan /Feb~<~~~~Povey, Jennifer
R.~
Each approach has its advantages and disadvantages. The original format presents the data in a much more "formal" fashion which allows for validation and data mining and ensures a consistent layout in the index; however, it does not distinguish between authors, editors, translators and/or subjects of the book being reviewed.
An attempt was in August 2015 (thread "Book Reviews") to reconcile the two approaches but without success, but the v1 indexing programs were modified in August 2016 to match the two different styles (e.g. the two reviews of Modern Science Fiction above which are the same review) so there was no real need to resolve the incompatibility. Currently the v2 indexing programs do not match the two styles.
In August 2015, while indexing Harper's, Ken Johnson started listing engravers separately from artists and requested that these be supported in the index. A "quick and dirty" solution was identified at the time by adding ", engraver" to the artist name and then using an 04 record in PSEUD.CVT to normalise the name to the form without the suffix - this would mean that the suffix would be listed in the Issue Index for Harper's but have no effect elsewhere. It was agreed at the time that this was a stopgap measure and that support for a formal author qualifier of " ,engraver" would be better, but needed further thought before implementation.
The subject resurfaced in August 2016 and after much (at times heated) discussion it was agreed that distinguishing engravers with a formal author qualifier was actually not a viable solution for three key reasons:
As such it was agreed (13-Aug-2016, "Re: Engravers (again)") not to provide formal support for a " ,engraver" author qualifier, although the "quick and dirty" solution of using a name suffix of ", engraver" could be continued if it was felt appropriate. (Closed)
The current approach whereby all series are independent of each other is somewhat clumsy. For example William Hazlett Upson wrote a long series of stories about Alexander Botts, a salesman for Earthworm Tractors. However, he also wrote some earlier stories about Earthworm Tractors that did not include Alexander Botts. At the moment the only way of handling this is to have two separate series for "Botts| Alexander" and "Earthworm Tractors" and to list every Botts story as being in both series which is both unnecessary additional work and looks very clumsy in the index.
It would be preferable to have some way (ideally nested) of indicating that the "Alexander Botts" series is a subset of the Earthworm Tractors series so that only a single series entry is needed in the data (and in the author/contents indexes) while the series index would contain something like:
This could be handled in SERIES.CVT by means of a special character in the third field, as in:
Earthworm Tractors~0~Upson, William Hazlett~
Botts| Alexander~0~@Earthworm Tractors~
This was discussed briefly in January 2017 but Bill C. decided it would be too difficult to implement with the current programs so the idea was shelved.