Hello Lists,
I think I finally managed to put the development version of pdfreaders in working order. Now I need your assistance to test the page on different platforms and devices, to prepare translations and to collect and refine suggestions for pdf reader software.
The working page is: http://pdfr.plutonium.fsfeurope.org and we are planning to switch it to the productive domain BY THE END OF THE WEEK.
The SVN branch is: https://svn.fsfe.org/pdfreaders/branches/2013
Here is how you can help:
== Add readers and refine reader information == What's to be done? - Some of the readers on the the list may not be maintained anymore. Find out which, and see to it, that they get kicked out, or marked accordingly. - Some of the readers don't have a working link to the project homepage yet, find the homepage and put it on record. - Some of the readers have incomplete or wrong information regarding platform support and browser integration. Find out which platforms are actually supported by a given reader and complete the list. - some contemporary readers are not yet on the list. Inform us about those, or better yet add them yourself.
How to do it? Readers are defined by json files - a generic file format which is slightly easier for humans to handle than xml. The reader files can be found here: https://svn.fsfe.org/pdfreaders/branches/2013/readers/ Each reader file contains basic information about the reader, a list of operating systems on which the reader runs (the link to an installer is assigned to the OS section), a list of web browsers for which a reader provides plugins, and a list of languages in which the reader is available. Adapt an existing reader file or create a new one and upload it to the svn repository, or send it to this mailing list, so someone else can upload it. Evince and Okular are the most completely defined readers and their definition files should provide good examples. Also see the description of the selection heuristics at the end of this email.
== Adapt translations == What's to be done? For the most part translations can be reused from the old page. However the file path is slightly different (each page got its own sub directory now) and some paragraphs in the "index" and "readers" page changed. In particular there is now some placeholder sections on these pages and the file readers/reader.??.snip has been created to hold some single words for translation. Some of this work can even be done without knowledge of the languages involved.
How to do it? Look at the file hirarchies of the old site: https://svn.fsfe.org/pdfreaders/trunk/src/ and the new site: https://svn.fsfe.org/pdfreaders/branches/2013/src/ Note that all the page files were renamed to index.??.page and moved to a sub directory corresponding to their old name. Note that this might affect hyper links contained in the pages. If you have access to the svn repository, see to it, that all language files get copied to the new position and check hyper links. If you can adapt the translation in one of the reader or index files, please do so and update the file or mail it to the list. The same goes for the new snippet file (reader.??.snip).
== Test the page == What's to be done? - Check hyper links on all translated pages. - Check reader suggestions on different platforms, with different browser configuration. - Whatever you come up with. - Report problems to this list.
How to do it? Get access to different Operating Systems and web browsers, including proprietary software setups. See if the page appears sensible with all setups. Hyperlinks are not setup dependent, except links to installers. For whitebox testing take a look into: https://svn.fsfe.org/pdfreaders/branches/2013/src/scripts/
------------------------------------------------------------------------ == Technical: How is reader selection done? == -- also: what can still be improved --
First thing we need to do is parse the user agent string of a web browser. From this we can identify the OS and browser of the visitor. Unfortunately the agent strings are pretty chaotic and all ready made software libraries we came up with failed in one way or another. So I wrote a new one... which is going to need occasional maintenance when new platforms appear in the wild. We detect the operating system and browser by matching the entire UA string against a sequence of regular expressions. The order of those regexes is important, since many systems include identifiers of other systems in their agent strings. On the first match we return a simplified system / browser identifier to which we refer as "canonical_name". The detection module contains the list of regexes and canonical names: https://svn.fsfe.org/pdfreaders/branches/2013/src/scripts/modules/AgentInfo....
At startup we read all the reader definition files and press them into useful data structures. That is: - One Hash containing all canonical OS names as keys. The value to each key is an array listing all readers that can be run on the corresponding OS. - One Hash containing all canonical browser names as keys. The value to each key is an array listing all readers that can be embedded in the corresponding browser.
Once we have identified the canonical name of the page visitors OS and browser, we just need to fetch the list with corresponding readers from those hashes. Thus we end up with an array, listing all sensible readers for the OS and a second array, listing all sensible readers for the browser.
Ideally we want to suggest a reader, which runs on the visitors platform and does also integrate with her browser. So we calculate the intersection of both lists and end up with a short list of all readers that fit this condition. This list may be empty, in which case we choose to continue working with only the OS based list. All readers have a priority assigned to them in the definition file. This is because we would rather recommend the fancy and fashionable readers (Evince, Okular) over the exotic (BePDF) and ugly (Motif/Xaw3D based) ones. So as a last step we sort the list by priority (lowest number is best number) and suggest the top most reader on the front page. The readers page contains the entire list of recommendable readers plus the list of all readers which won't work well with the visitors setup.
Most part of the reader selection is done in: https://svn.fsfe.org/pdfreaders/branches/2013/src/scripts/modules/Suggest.pm
The files index.pl and readers.pl are the main scripts that deliver the only two dynamic pages on the site. They can be viewed here: https://svn.fsfe.org/pdfreaders/branches/2013/src/scripts/
The language to which a reader has been translated does currently *not* affect the reader selection. Even though, the definition files list the software languages. All reader definition files are read at script startup, which currently means on each page impression. By porting the scripts to FastCGI/WSGI there is a *lot* of time to be saved. ---------------------------------------------------------------------- End Technical Notes
Thank you for your patience, for all the suggestions I didn't embrace yet, and for all the work you have put into pdfreaders.org so far. Now is the time, to introduce new reader software and to get around to giving the new site it's finish.
Sincerely,
Dear *,
On 02/06/13 23:36, Paul Hänsch wrote:
I think I finally managed to put the development version of pdfreaders in working order. Now I need your assistance to test the page on different platforms and devices, to prepare translations and to collect and refine suggestions for pdf reader software.
Excellent. Good work.
Some thoughts follow…
1) I will try to provide a translation guide for PDFreaders.org by the end of the week. we should pronbably link to it from the footer, and it should fit in without any problems.
2) The text on pdfreaders.org has been justify-aligned for some time, and I would like to see that continue. Unless anyone objects, I will update the CSS on Paul's branch.
3) Philipp, you provided awesome updates to the icon (in this list as "PDFreaders.org icon's suitability as a MIME type icon for PDF files"). Can we come up with an installation script, add a CC0 licence, and get the updated graphics online with the new site? It seems especially important because the icon looks horrible in the readers list. :-/
4) There are some mobile readers in the Wiki and on Trac that should be checked out and possibly added to the website. Does anyone on the list have time for this? Also, as Paul mentioned, some readers are no longer actively developed and should be removed.
Cheers,
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Dear all,
On 03/06/13 03:10, Heiki "Repentinus" Ojasild wrote:
- I will try to provide a translation guide for PDFreaders.org by
the end of the week. we should pronbably link to it from the footer, and it should fit in without any problems.
I have produced a contribution guide. At the moment this resides on the test branch as contribute.en.page. I also added some sections to the default.css, in adddition it has its own CSS file toc.css.
You can see the current version at http://test.pdfreaders.org/contribute.en.html. You are welcome to provide recommendations on page style, text wording, &c.
I do not wish to see it in the main horizontal menu; however, as "in_menu: false;" seems to have no effect, I do not know how to resolve that. Any ideas? I would like to link to it from the footer if no-one objects.
(Ignore the "Icon Initiative" branch on test, this still needs work.)
When we can get it off the main menu and I have edited the footer, I would like to move this to Paul's branch (Paul, OK?), so we can get this live alongside the other updates.
Cheers, - -- Heiki "Repentinus" Ojasild FSFE Fellowship Representative mailto:repentinus@fsfe.org xmpp:repentinus@jabber.fsfe.org http://blogs.fsfe.org/repentinus/
On 06/03/2013 03:36 AM, Paul Hänsch wrote:
The working page is: http://pdfr.plutonium.fsfeurope.org and we are planning to switch it to the productive domain BY THE END OF THE WEEK.
Probably I'm doing something wrong.. it doesn't work with my default browser; wget returns a file of zero size.
Ineiev ineiev@gnu.org, Mon, 03 Jun 2013 07:22:37 +0400:
On 06/03/2013 03:36 AM, Paul Hänsch wrote:
The working page is: http://pdfr.plutonium.fsfeurope.org and we are planning to switch it to the productive domain BY THE END OF THE WEEK.
Probably I'm doing something wrong.. it doesn't work with my default browser; wget returns a file of zero size.
Thanks, a faulty regex in the browser detection for Links and ELinks, blocked all subsequent regex checks.
Reminds me, we should add pdftotext or similar as a reader for terminal users.
For now, I've made xpdf the reader which gets suggested if no known platform was recognised. Noticed that the install URLs there are nonsensical anyway.
We should think of a better default / last resort too.
What was your default browser anyway? One of the text-browsers or a non-Firefox Gecko based one?
On 06/03/2013 09:45 AM, Paul Hänsch wrote:
Ineiev ineiev@gnu.org, Mon, 03 Jun 2013 07:22:37 +0400: Thanks, a faulty regex in the browser detection for Links and ELinks, blocked all subsequent regex checks.
Now it basically works.
For now, I've made xpdf the reader which gets suggested if no known platform was recognised. Noticed that the install URLs there are nonsensical anyway.
What about seeing options for other platforms or configurations? I can imagine situations when the visitor wants to use the reader outside of her current browser, or (to suggest somebody) to use it with a different browser or under a different OS.
What was your default browser anyway? One of the text-browsers or a non-Firefox Gecko based one?
Yes, it is a text browser.
Thanks for your work first of all.
On 03.06.2013 09:11, Ineiev wrote:
On 06/03/2013 09:45 AM, Paul Hänsch wrote:
Ineiev ineiev@gnu.org, Mon, 03 Jun 2013 07:22:37 +0400: Thanks, a faulty regex in the browser detection for Links and ELinks, blocked all subsequent regex checks.
Now it basically works.
I still see only an empty page. Chromium on FreeBSD, here.
Don't you think after this long development time, we should give the site more than a week of testing?
Hannes Hauswedell h2@fsfe.org, Mon, 03 Jun 2013 22:05:08 +0200:
I still see only an empty page. Chromium on FreeBSD, here.
Yeah. Happened when your platform was actually recognised correctly, but no registered readers for this platform where available.
I've implemented a fallback, which redirects you to the full readers list now in this case.
Don't you think after this long development time, we should give the site more than a week of testing?
No, I really don't. I don't think there are that many actual *bugs*. The two we found were easily recognised and easily fixed.
The much bigger problem here is the lack of reader data. That was the first list item in my original mail. People who are visiting the page now might see Evince or Okular suggested to them, and think the data is fine. But it is really not.
A lot of the platform listings and and project links are simply fantasy data. Most readers are not checked, not reviewed or anything, the information on the site is pure test data, nothing more! This affects the proper selection of readers to suggest and the simplicity of the reader installation.
I cannot change this myself. If we wait another month, I still cannot fix it and the project is already three month over its deadline. If we can't accumulate the capacity to fill in those gaps on our own, then we will simply have to rely on the "outside" world to help us out. For this the site should be online rather soon than later.
That doesn't actually mean we have to advertise it yet to governments and organisations. But at least helpful people must be able to see our need.
The most important thing for now is, that someone who would otherwise just klick the download button for the adobe reader will be presented with something, that works about as well. Everything else must be filled in by people who are interested in the site and friendly to our cause.
So I'd like to urge you all again, to upload or send reader definition files.
Thank you
ew snippet file (reader.??.snip).
== Test the page == What's to be done?
- Check hyper links on all translated pages.
- Check reader suggestions on different platforms, with different browser configuration.
- Whatever you come up with.
- Report problems to this list.
How to do it? Get access to different Operating Systems and web browsers, including proprietary software setups. See if the page appears sensible with all setups. Hyperlinks are not setup dependent, except links to installers. For whitebox testing take a look into: https://svn.fsfe.org/pdfreaders/branches/2013/src/scripts/
From my office it seems to work, some notes on the layout: * Spacing between the icon and the surrounding dashed box is too small * text in dashed box is not vertically aligned * use "Platform" word instead of "system" * spacing above and below dashed box too small, especially on Readers-page * license is irrelevant to the user, it should suffice that it is Free Software, at least on the front page * Download link should be direct link to latest installer * in addition to Download link, there should be a homepage link IMHO
Thanks for your work already!
Best, Hannes
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 02/06/13 23:36, Paul Hänsch wrote:
== Add readers and refine reader information == What's to be done?
- some contemporary readers are not yet on the list. Inform us
about those, or better yet add them yourself.
I just committed a .json for pdf.js. The json schema enum list does not contain Android or Windows 8. Since Android had already been used, I added this. I commented out the entry for Windows 8 until you can resolve this issue.
Furthermore, regarding browsers, I do not see a point in having to provide a maximum version. This seems like a huge amount of future maintenance; hence, I recommend making it possible to specify a minimum version only.
By the way, is this branch supposed to rebuild automatically? It does not seem to do so…
Also, as I have commented before, the fallback icon looked terrible. I provided a 100px² PNG from the SVG and updated all symlinks.
Cheers, - -- Heiki "Repentinus" Ojasild FSFE Fellowship Representative mailto:repentinus@fsfe.org xmpp:repentinus@jabber.fsfe.org http://blogs.fsfe.org/repentinus/