Discussion:
Referencer 1.0.4-pre
John Spray
2007-05-30 22:16:25 UTC
Permalink
Hi,

The "Changelog longer than index finger" rule demands that I make
another release :-)

There have been a few relatively invasive changes this time around, so I
would appreciate if one or two people could check that this tarball
compiles and runs before I unleash it on the unsuspecting public.

Cheers,
John
-------------- next part --------------
A non-text attachment was scrubbed...
Name: referencer-1.0.4-pre.tar.gz
Type: application/x-compressed-tar
Size: 400965 bytes
Desc: not available
URL: <http://icculus.org/pipermail/referencer/attachments/20070530/ac09fd0f/attachment.bin>
Tom Purnell
2007-05-30 22:40:17 UTC
Permalink
On Wed, 30 May 2007 23:16:25 +0100
Post by John Spray
Hi,
The "Changelog longer than index finger" rule demands that I make
another release :-)
There have been a few relatively invasive changes this time around,
so I would appreciate if one or two people could check that this
tarball compiles and runs before I unleash it on the unsuspecting
public.
Cheers,
John
So far so good here,

This is the first version that has allowed me to save a library without
FUBARing on my amd64 machine

Finally I can put this great program to real use :)
--
Tom Purnell, Est 1983
http://www.thomaspurnell.com
Rodrigo Kassick
2007-05-31 02:02:22 UTC
Permalink
Compiles fine here (ubuntu 32 bits)
Runs ok; no crashers till now
Post by Tom Purnell
On Wed, 30 May 2007 23:16:25 +0100
Post by John Spray
Hi,
The "Changelog longer than index finger" rule demands that I make
another release :-)
There have been a few relatively invasive changes this time around,
so I would appreciate if one or two people could check that this
tarball compiles and runs before I unleash it on the unsuspecting
public.
Cheers,
John
So far so good here,
This is the first version that has allowed me to save a library without
FUBARing on my amd64 machine
Finally I can put this great program to real use :)
--
Tom Purnell, Est 1983
http://www.thomaspurnell.com
--
Rodrigo Virote Kassick
(k?zic/paiacan)
------------------------------------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://icculus.org/pipermail/referencer/attachments/20070530/91c8fa63/attachment.htm>
Rick L Vinyard Jr
2007-05-31 03:09:15 UTC
Permalink
Post by John Spray
Hi,
The "Changelog longer than index finger" rule demands that I make
another release :-)
There have been a few relatively invasive changes this time around, so I
would appreciate if one or two people could check that this tarball
compiles and runs before I unleash it on the unsuspecting public.
Cheers,
John
Compiles and runs great on Fedora development x86_64. I love the new
features like the variable sized fonts on the tag groups, and the tool
tips with the reference summary!!!

But, a few items I noted:
- In list view the Title column can't be shrunk... only expanded

- It would be nice if the list view was in a Gtk::ScrolledWindow

- It would also be nice if there were icons on the toolbar to switch
between the icon view and list view

---

Rick L. Vinyard, Jr.
John Spray
2007-05-31 07:26:42 UTC
Permalink
Post by Rick L Vinyard Jr
- In list view the Title column can't be shrunk... only expanded
Yup, the list view needs some love, it's on my mental TODO list for the
next release.
Post by Rick L Vinyard Jr
- It would be nice if the list view was in a Gtk::ScrolledWindow
It is, but it's set to only shown the vertical scrollbar when necessary,
and never to show the horizontal one. If I let it show the horizontal
one it tends to always show it, and the width is determined by the
longest title+author.

On the TODO for 1.0.5 are:
- Sort out the column sizing
- Make the fields editable without opening document properties
- (Maybe) Look into adding/removing columns
- Save the sorting state between sessions

Anyone have anything to add?
Post by Rick L Vinyard Jr
- It would also be nice if there were icons on the toolbar to switch
between the icon view and list view
I thought about doing that, but guessed that people generally use one or
the other rather than switching often. Anyone else found themselves
wishing for this?

John
Rick L Vinyard Jr
2007-05-31 16:52:26 UTC
Permalink
Post by John Spray
- Sort out the column sizing
- Make the fields editable without opening document properties
- (Maybe) Look into adding/removing columns
- Save the sorting state between sessions
Anyone have anything to add?
Just a few personal wishlist items; and before I list them... I'll help
write them if you want.

* Hierarchical tags... i.e. tags organized by trees.

* Paste BibTeX in the document properties dialog

* BibTeX fields that change based on the type in the document
properties dialog
http://en.wikipedia.org/wiki/Bibtex seems to have a pretty good list of
what is required and what is optional for each type.

* On the document right-click menu, in addition to the 'Open...'
option, an 'Open With...' option.

* Per document notes... i.e. a mechanism for associating notes with a
particular item.
Perhaps a simple way to do this would be to leverage existing notetaking
apps (Zim, Tomboy, gnotes). Looking at Zim in particular it takes a
parameter that is the notes page. With an additional string stored in
BibData, a user specified notes app (probably set in preferences) and a
menu entry, an execle() could spawn the notes app with the associated
string as the parameter.

* Related to the per document notes above... have a mechanism for
opening an associated xournal file.
I've been using xournal to annotate PDFs, and referencer doesn't have
any problem importing them. But, it would be nice if the xournal file
were associated with the actual paper.
Post by John Spray
Post by Rick L Vinyard Jr
- It would also be nice if there were icons on the toolbar to switch
between the icon view and list view
I thought about doing that, but guessed that people generally use one or
the other rather than switching often. Anyone else found themselves
wishing for this?
Personally, I like to use the icon view the most and the tooltips make
it even better. But, every now and then I find myself switching to the
list view when I want to scan everything in a tag.
Frederik Elwert
2007-05-31 17:25:44 UTC
Permalink
Post by Rick L Vinyard Jr
Just a few personal wishlist items; and before I list them... I'll help
write them if you want.
* Hierarchical tags... i.e. tags organized by trees.
Right, I proposed something related. But I'd prefer ad-hoc trees based
on the occurrence of related Tags over fixed hierarchies. What do you
think about this?
Post by Rick L Vinyard Jr
* BibTeX fields that change based on the type in the document
properties dialog
http://en.wikipedia.org/wiki/Bibtex seems to have a pretty good list of
what is required and what is optional for each type.
Oh, this would be so cool! Then referencer would turn into a more
general purpose bibliography tool and not just an article manager. Now,
I still use pybliographer for my BibTeX files and have a second database
for my PDF articles in referencer. Using only one tool would be really
nice.
Post by Rick L Vinyard Jr
* Per document notes... i.e. a mechanism for associating notes with a
particular item.
Besides the external-app-solution, referencer could have it's own way of
managing notes - there is a corresponding BibTeX-field, so it would make
sense in a way.

I love to see how actively referencer's future is shaped...

Cheers,
Frederik
John Spray
2007-05-31 17:45:50 UTC
Permalink
Post by Frederik Elwert
Post by Rick L Vinyard Jr
Just a few personal wishlist items; and before I list them... I'll help
write them if you want.
* Hierarchical tags... i.e. tags organized by trees.
Right, I proposed something related. But I'd prefer ad-hoc trees based
on the occurrence of related Tags over fixed hierarchies. What do you
think about this?
I guess the main difference between actual trees and an automatic
tree-like display is that if the trees were specified by the user then
he could have a situation where when he has

Mammal
- Dog
- Cat
Fish
- Shark
- Goldfish

Then when he tags a file as "Shark" he gets the "Fish" tag
automatically, rather than having to specify it by hand.

Personally I don't have a preference, since I use a flat list of tags
anyway.
Post by Frederik Elwert
Post by Rick L Vinyard Jr
* BibTeX fields that change based on the type in the document
properties dialog
http://en.wikipedia.org/wiki/Bibtex seems to have a pretty good list of
what is required and what is optional for each type.
Oh, this would be so cool! Then referencer would turn into a more
general purpose bibliography tool and not just an article manager. Now,
I still use pybliographer for my BibTeX files and have a second database
for my PDF articles in referencer. Using only one tool would be really
nice.
Evil pybliographer user ;-)

Is the knowing which fields for which doc type the main thing that keeps
you with pybliographer or is it other things? I notice they have
medline search which we don't have (yet).
Post by Frederik Elwert
Post by Rick L Vinyard Jr
* Per document notes... i.e. a mechanism for associating notes with a
particular item.
Besides the external-app-solution, referencer could have it's own way of
managing notes - there is a corresponding BibTeX-field, so it would make
sense in a way.
The note field is used more often for things like @Unpublished entries
with a \url, right? Not sure it would be a good idea to populate it
with annotation-type notes.

John
Frederik Elwert
2007-05-31 22:05:46 UTC
Permalink
Post by John Spray
I guess the main difference between actual trees and an automatic
tree-like display is that if the trees were specified by the user then
he could have a situation where when he has
Mammal
- Dog
- Cat
Fish
- Shark
- Goldfish
Then when he tags a file as "Shark" he gets the "Fish" tag
automatically, rather than having to specify it by hand.
Yes, right, it also has it's benefits. But in my opinion, the automatic
tree display doesn't prevent this kind of things. If shark always
implies fish, then it's sufficient to just select shark. If you want to
have a look what you have about fish, and then refine your selection to
only match shark, then it's also possible. But your first-order tag list
gets longer if you don't have a pre-defined hierarchy, that may be a
disadvantage.
Post by John Spray
Evil pybliographer user ;-)
Hm, I don't think so. Pybliographer has some great possibilities, and
their development branch may become a first-class framework for managing
bibliographies in arbitrary formats. Maybe even referencer might benefit
from this one day, if support for more formats is planned. But currently
their GUI is lacking some love, and I'm just waiting for tag support for
too long. Now I can either hack it in for myself (which I might be able
to do some day, as I have at least some python knowledge), or I switch
to some promising project that already has this kind of great stuff (but
which I sadly can't help by coding, as I have no clue about C++) :-)
Post by John Spray
Is the knowing which fields for which doc type the main thing that keeps
you with pybliographer or is it other things? I notice they have
medline search which we don't have (yet).
I don't use medline. Z39.50 would be nice... ;-)
No, it's mainly the field selection. Manually add the necessary fields
is a bit odd. And one thing I really like is the automatic key
generation. You just fill in author and year, and it generates a key. So
"Henry Jones, 2003" will generate "Jon03", and if this already exists,
"Jon03b". That's nice, as keys as unique identifiers have a rather
technical meaning.
Post by John Spray
with a \url, right? Not sure it would be a good idea to populate it
with annotation-type notes.
As I don't know all the details about BibTeX, I often take
http://www.ecst.csuchico.edu/~jacobsd/bib/formats/bibtex.html
as a reference. It distinguishes a note field for information that is
relevant in the bibliographic output, and an annote field that is not
used by the standard styles. Maybe it would be ok to use this one.
And for URLs, there is a (non-standard, but widely used) URL field.

But it might not be necessary to use a BibTeX field for annotations,
referencer can of course use it's own field for this. Or use external
applications. I don't know which might be the best way. Personally, I
use zim for longer abstracts that include formatting, so configurable
external programs would be fine with me... :-)

Cheers,
Frederik

John Spray
2007-05-31 17:32:57 UTC
Permalink
I'll help write them if you want.
Brave man, consider yourself volunteered :-)
* Hierarchical tags... i.e. tags organized by trees.
If you can figure out a way to do it without complicating the UI. I
would impose the condition that the existing UI shouldn't become any
more complicated, and people who don't want trees of tags shouldn't be
exposed to any fallout.
* Paste BibTeX in the document properties dialog
Would be useful for the "add a pdf with some bibtex" use case. However,
why put it in the dialog rather than having a "Paste into" operation in
the main window?
* BibTeX fields that change based on the type in the document
properties dialog
http://en.wikipedia.org/wiki/Bibtex seems to have a pretty good list of
what is required and what is optional for each type.
This will require a considerable bit of careful restructuring behind the
scenes. Although such restructuring is fine by me, please bear in mind
that Referencer is not intended to be a bibtex-specific tool.
Constraints on which fields are valid for which document types may vary
between bibliography formats. As such, all this stuff needs to be
designed such that information about which fields are visible with which
document types is dynamic, and takes into consideration future export
targets such as MS Word bibliography files and OpenOffice.

Disclaimer: the bibdata stuff was written waaay before referencer was
ever going to be publically available! As such it's "nobody's ever
going to look at this" code.

Also, the document properties dialog needs to stay neat. The fields
should always be presented in a reliable order, and things that only
need half-width fields (such as year, issue # etc) should be presented
as such. I would like to avoid a list of fields so long it's scrollable
as far as humanly possible, possibly maintaining some fields as visible
by default, and others in a "More Fields" expander which may be valid
for a document type but are not so frequently used.
* On the document right-click menu, in addition to the 'Open...'
option, an 'Open With...' option.
My own use case for this is that I occasionally want to open a pdf with
acroread instead of evince -- what's yours?
* Per document notes... i.e. a mechanism for associating notes with a
particular item.
Perhaps a simple way to do this would be to leverage existing notetaking
apps (Zim, Tomboy, gnotes). Looking at Zim in particular it takes a
parameter that is the notes page. With an additional string stored in
BibData, a user specified notes app (probably set in preferences) and a
menu entry, an execle() could spawn the notes app with the associated
string as the parameter.
Some kind of layer that would allow referencer to interact with various
note-taking frameworks would be desirable, although support for a single
program would be acceptable as a prototype.

At this point we start to enter somewhat murky territory where what we
really want is a plugin system. A relatively big job.
* Related to the per document notes above... have a mechanism for
opening an associated xournal file.
I've been using xournal to annotate PDFs, and referencer doesn't have
any problem importing them. But, it would be nice if the xournal file
were associated with the actual paper.
I guess really what you're suggesting here is being able to associate
more than one file to a reference. Seems pretty viable, through a
concept of "Attachments" or so.

If you're serious about getting your hands dirty with this, feel free to
get me off-list on IRC (#referencer on freenode.net) or google talk
(username jcspray) to discuss code details, including telling me I'm
wrong about how things should be done ;-)

Although I wouldn't encourage using it as a wishlist in general, if
you're thinking about implementing something you might want to write up
a quick spec as a "Blueprint" on launchpad:
https://blueprints.launchpad.net/referencer/

John
Brice Goglin
2007-05-31 17:56:36 UTC
Permalink
Post by John Spray
- Sort out the column sizing
- Make the fields editable without opening document properties
- (Maybe) Look into adding/removing columns
- Save the sorting state between sessions
Anyone have anything to add?
I would still like to have referencer's tags exported to in a special
field in bibtex files for external filtering later, as explained earlier in

http://icculus.org/cgi-bin/ezmlm/ezmlm-cgi?60:mss:36:jiifbeijbklioocmnlpo

Brice
John Spray
2007-05-31 18:00:20 UTC
Permalink
Post by Brice Goglin
Post by John Spray
- Sort out the column sizing
- Make the fields editable without opening document properties
- (Maybe) Look into adding/removing columns
- Save the sorting state between sessions
Anyone have anything to add?
I would still like to have referencer's tags exported to in a special
field in bibtex files for external filtering later, as explained earlier in
Tags are currently getting exported as the "keywords" field. I didn't
fully understand from your previous message whether (or why) you
specifically needed them in a field with a different name?

John
Brice Goglin
2007-05-31 18:29:54 UTC
Permalink
Post by John Spray
Post by Brice Goglin
I would still like to have referencer's tags exported to in a special
field in bibtex files for external filtering later, as explained earlier in
Tags are currently getting exported as the "keywords" field.
Oh great, I didn't see this, thanks a lot!
Post by John Spray
I didn't
fully understand from your previous message whether (or why) you
specifically needed them in a field with a different name?
It was just a suggestion, since it could conflict with something else
already using the "keyword" field. Prefixing with "referencer" would
avoid conflicts, but I am not sure it is worth doing so. A configurable
prefix would be great, but it's not high-priority for me.

Thanks again, 1.0.4 seem to be ready for intensive usage here.
Brice
Brice Goglin
2007-05-31 06:17:22 UTC
Permalink
Post by John Spray
Hi,
The "Changelog longer than index finger" rule demands that I make
another release :-)
There have been a few relatively invasive changes this time around, so I
would appreciate if one or two people could check that this tarball
compiles and runs before I unleash it on the unsuspecting public.
Cheers,
John
No problem here, and I can even import my bibtex file in a non-UTF8
environment, great!

You might want to #include <cmath> in TagWindow.c to avoid a warning
about logf().

Thanks,
Brice
Loading...