Bernard,

the command does not refer
> to a meta-repository but does not work better !


You need to append /trunk to the URL:

$ svn list -v http://github.com/apertium/apertium-fra-por/trunk

provides the same output as SVN does. This is by design so that you can
access branches. If our SVN structure were less non-standard, all the
directories
would be under trunk.

For released pairs, there is a website doing on-line translations.


There is also an (official?) site that offers all the pairs:
http://beta.apertium.org/
along with analysis and generation functionality.

Since it depends on apertium-get which has not been updated yet either,
it is also a bit out-of-date. Hopefully, apertium-get can be updated soon
for Git.
There is an issue open for it (I may tackle it at some point soon).

To correct this problem, the only need is a command giving for each branch
> the list of the pairs inside with the DATE (not n days / months / years
> ago)
> of the last change. Exactly what svn list -v does.


This can be done with a few commands

1) download the .gitmodules file in each meta repository with wget from the
raw
GitHub blob (or svn export I think).
2) loop through the .gitmodules file and svn ls for each repository.

The script itself would take more time but the task can be accomplished
relatively
simply in any language. Step 1 can also be done via the GitHub API, which
the JS
and Bash in the apertium-on-github repo do (the API might also give you
some of
the information you desire, replacing Step 2).

Javascript is not a solution as it is executed by a web browser.


Node.JS is an extremely well supported platform with probably as much
documentation as Java or Python that enables executing JS outside of a
browser
environment. Regardless, Bash equivalents (relying on Python for JSON
parsing
but jq can be substituted for it) and Python, are also present.

This repository just gives few uncommented scripts without explaining
> how to write them.


I'm happy to comment anything non-trivial, Please leave a comment through
GitHub
on any code that you would like explained in-line.

Well, I have another approach. If the github Apertium repository is the
> reference for Apertium, it should include all what has been developed
> for Apertium.


That's a valid perspective. There were no concerns with this approach when
the proposal
was put forward. Any member of the Apertium org on GitHub can feel free to
migrate tools
at their convenience using the well-documented scripts and instructions in
the migration
repository. Figuring out where tools begin and end, where they belong and
how they should
look is often a non-trivial process especially when almost all of them are
missing active
maintainers.

According to the idea of putting somewhere else tools not under
> development, why not then having a complete reference to the whole
> Apertium project.


Something like this would be nice to have. I'm happy to guide with regards
to the best way
of fetching information whether it be from GitHub APIs, the SVN bridge or
Git itself.

but a change should be done in the file
> desmnemo.c of apertium-c-formatters

.


Could you confirm this means you want the apertium-c-formatters directory
in SVN migrated
to a GitHub repository?


[image: Sushain Cherivirala]
*Sushain K. Cherivirala *
Stanford University, M.S. in Computer Science '19
Carnegie Mellon University, B.S. in Computer Science '18
(713) 992-4043  <713-992-4043>| www.skc.name

On Sat, Apr 7, 2018 at 4:16 AM, Bernard Chardonneau <bechapert...@free.fr>
wrote:

> > Date: Mon, 2 Apr 2018 14:02:46 -0500
> > From: Sushain Cherivirala <sush...@skc.name>
> > To: Bernard Chardonneau <bechapert...@free.fr>
> > Cc: Francis Tyers <fty...@prompsit.com>, "Mikel L. Forcada" <
> m...@dlsi.ua.es>,
> >         Shardul Chiplunkar <shardul.chiplun...@gmail.com>, Tino
> Didriksen
> > <tino.didrik...@gmail.com>,
> >         Jonathan Washington <jonathan.n.washing...@gmail.com>, Xavi
> Ivars
> > <x...@infobenissa.com>,
> >         "[apertium-stuff]" <apertium-stuff@lists.sourceforge.net>
> > Subject: Re: Command line for listing Apertium branches and other Git
> problems
>
> > Hi Bernard,
>
> Hi
>
> >> Last February Sushain indicated me a command-line to list Apertium
> >> branches and files, but it does not work with this git repository :
> >>
> >> $ svn ls -v http://github.com/apertium/apertium-languages
> >> Redirecting to URL 'https://github.com/apertium/apertium-languages':
> >>     287 apertium              Apr 02 12:31 ./
> >>       1 apertium              Mar 08 03:42 branches/
> >>     287 apertium              Apr 02 12:31 trunk/
> >> $ svn ls -v http://github.com/apertium/apertium-nursery
> >> Redirecting to URL 'https://github.com/apertium/apertium-nursery':
> >>      59 apertium              Mar 28 09:16 ./
> >>       1 apertium              Mar 08 03:42 branches/
> >>      59 apertium              Mar 28 09:16 trunk/
> >> $ svn ls -v http://github.com/apertium/apertium-nursery/branches
> >> Redirecting to URL 'https://github.com/apertium/
> apertium-nursery/branches
> >> ':
> >>      59 apertium              Mar 28 09:16 ./
> >> $ svn ls -v http://github.com/apertium/apertium-nursery/trunk
> >> Redirecting to URL 'https://github.com/apertium/apertium-nursery/trunk
> ':
> >>      59 apertium              Mar 28 09:16 ./
> >>      39 apertium         5062 Mar 15 13:12 .gitmodules
> >>       1 apertium           19 Mar 08 03:42 README.md
> >>
> >> Normally that should list languages or language pairs and give the date
> >> of their last change.
>
> > As I mentioned in my emails, the Git-SVN bridge will not work well for
> the
> > meta-repositories such as -tools, -incubator, etc.
>
> Well, I think what happens since Apertium was migrated to Git and github
> is exactly the same kind of things than when political people ask to be
> elected on a program, and after been elected, don't follow this program
> and sometimes do the contrary.
>
> Of course, people who wanted to move to Git and github presented
> advantages.
>
> But one thing important is : if something works fine with subversion, there
> is no reason to accept it stops working with Git.
>
> There were two previous tentative to ask PMC to move to Git and github, and
> according what happens now, I think people who previously voted against
> this
> idea were right : Git has possible advantages but we did not know what we
> would loose.
> http://wiki.apertium.org/wiki/PMC_proposals/Move_apertium_to_github#Voting
> http://wiki.apertium.org/wiki/PMC_proposals/Allow_some_code_
> under_github.com/apertium#Voting
>
> So, some examples about what was (quite) done last February and that would
> still work for years if git migration was not done :
>
> On Apertium project, something like 291 language pairs have been started.
> Some of them reached a high quality level and were released, other not.
>
> For released pairs, there is a website doing on-line translations.
> But :
> 1) released pairs if just a small part of what we can get using Apertium,
> 2) if you add words to a released pair, it may take years before a new
>    release will translate these new words on the official Apertium website.
>
> I am not against a choice by people of Alicant university and Prompsit for
> this on-line translator to have a full control of the state of language
> pairs used on it.
>
> But it does not forbid other unofficial translators using Apertium.
> And now, there is at least one.
> http://apertiumtrad.tuxfamily.org/
> At least 2 problems are still to fix (one with constraint grammar), but if
> no migration to git had been done, this website would continue fetching
> changes on Apertium languages and language pairs, download and compile the
> new versions.
>
> Instead of that, this website may now spend years translating using
> language
> pairs as they were on the 8th of march 2018 instead of using the current
> versions.
> To correct this problem, the only need is a command giving for each branch
> the list of the pairs inside with the DATE (not n days / months / years
> ago)
> of the last change. Exactly what svn list -v does.
>
> To know about language pairs and Apertium dictionaries, there are also wiki
> pages that were daily updated twice a day up to the 4th October 2011.
>
> As this update disappeared, I wrote scripts that permitted me to update
> these
> wiki pages when I was thinking to run them and tho copy generated files in
> the wiki pages. But after several years, manual updates became less and
> less
> frequent.
>
> But as now, I have a website with a cron able to scan Apertium language
> pairs, it was possible to fetch data to generate (out of the wiki)
> this : http://apertiumtrad.tuxfamily.org/listepaires.php
> and this : http://apertiumtrad.tuxfamily.org/listedic.php?sel=cpl
> (click to validate).
>
> And for the official Apertium wiki, using the same data, there is this :
> https://apertiumtrad.tuxfamily.org/genlistep.php?lang=eng
> https://apertiumtrad.tuxfamily.org/genlistedic.php?lang=eng
>
> Source code got from these two links can be copied in the apertium wiki
> pages, and if somebody wanted to add a bot to fetch theses URL, wiki
> pages could be updated again automatically.
>
> But (again) until having command-line to see the actual state of git
> Apertium repository, the result of all this will continue to show how
> it was on the 8th of march 2018 !
>
> According to svn list command (as you or shardul said these is a svn
> bridge),
> if works with, for instance the github repository for hfst :
>
> svn list -v https://github.com/hfst/hfst.git
> svn list -v https://github.com/hfst/hfst.git/trunk
> svn list -v https://github.com/hfst/hfst.git/trunk/scripts
> ..
>
> all work and give the date of last change for files and directories, who
> did this last change, and for files, the size.
>
> You said it "will not work well for the meta-repositories" (which is a
> problem), but :
>
> svn list -v http://github.com/apertium/apertium-fra-por
>
> or for any other language, language pair, tool, the command does not refer
> to a meta-repository but does not work better !
>
>
> >> Of course, there could be a workaround to get information from a web
> page :
>
> I thought to that possibility without testing it but downloading a URL from
> github website does not give a useful result because github.com does not
> respect the initial www philosophy.
>
> The standard format for websites is HTML and, except for some special
> usages
> like on-line file editing or video chat, the normal way is to send HTML
> from
> a server, and the web browser just has to display it.
>
> For informations including only text (which is the case for browsing a
> repository), this internet standard is sufficient, and then :
> - any web browser even if very old will display the page,
> - fetching the page form a script will be easy.
>
> Instead of that, github.com pages give code to be executed by the web
> browser.
>
> So :
> - with a little old browser (like Firefox 31), some information may not be
>   available (github.com ask to avoid old browsers)
> - then using a newer browser, some more informations appears
> - then returning to the old browser, the information can bee see also. It
> may
>   be cached somewhere.
>
> But :
> - several days, date of the last change is n days / months / years ago
> - other days, it may be a real date.
>
> And the most important, if you watch the source code, lot of informations
> displayed by the web browser will not be inside, but there will be scripts
> instead. So, this kind of web page is perfectly useless to get information
> with an automatic tool.
>
> > I suggest just using the GitHub API, it is probably the easiest solution:
>
> > e.g. curl https://api.github.com/repos/apertium/apertium-cat
>
> This is a json format list of link. But for the example above, we don't
> see the list of files of apertium-cat with their size, the date of their
> last change etc...
>
> > There are examples of how to use the API in both ES2018 (JavaScript) and
> > Python in this repository:
>
> > https://github.com/apertium/apertium-on-github
>
> Javascript is not a solution as it is executed by a web browser.
> This repository just gives few uncommented scripts without explaining
> how to write them.
>
> >> Another wrong thing with the web interface : apertium-incubator is not
> >> shown on the top https://github.com/apertium (Pinned repositories).
>
> > This is intentional. Only up to 6 repositories can be pinned. Some
> > discussion of which those 6 should be has taken place on IRC.
>
> That is a less important problem, as, if github.com website does not
> allow users enough control, it will be possible to put somewhere else
> on the web, a page that will contain the full list of Apertium branches.
>
> > The custom web interface is a bit more complete:
>
> > https://apertium.github.io/apertium-on-github/source-browser.html
>
> But (again) it need to execute a script on a (not too old) web browser.
> So, this page will be useless to get the list of pairs, language and
> tools directly from it's source code.
>
> >> It was a good idea to create a branch apertium-tools distinct from
> >> apertium-trunk which is now only for language pairs, but
> apertium-dixtools
> >> apertium-c-formatters and apertium-mediawiki seem to have disappeared.
>
> > Per the emails, tools will be migrated on an ad-hoc basis since many are
> > no longer in use/development.
>
> Well, I have another approach. If the github Apertium repository is the
> reference for Apertium, it should include all what has been developed
> for Apertium.
>
> I don't think a tool needs to be in development to be useful.
>
> For example, the UNIX / GNU/Linux pwd command is near than 50 years old
> and it is not sure this command is updated at least once a year. But it
> does not prevent it to be useful.
>
> How can you determine an Apertium tool is in USE ?
>
> For apertium-dixtools, I don't use it for sorting a dictionary. Other
> people did it, but I don't like having separate lists for adjectives,
> verbs, names etc... and so, not localising easily a word using the
> scrollbar to find where are the words starting by the same letters.
> So, I have my own (unpublished) sorting scripts.
>
> But there are other tools like merging or crossing dictionaries.
>
> As Apertium project started with separate language pairs to now prefer
> using reference monodices, a work will be to merge local monodices
> into the reference one to avoid unknown words when switching to the
> reference files.
> Even if for the whole Apertium project, this work is not done at least
> once every month, it's sure it will have to be done in the future.
>
> For apertium-c-formatters I used it a lot last year and this year
> (for mnemonic files). As there are sometimes interactions between
> two following lines of a mnemonic file, I found a simple idea to
> avoid them. But as now Apertium moved to Github, the normal way to
> work will be to first download a git included version (even if the
> most important files are already on my computers), and then to do the
> change.
>
> As it disappeared from apertium-tools after being there around the
> 8 of march, a possibility would be for me to rebuild the repository
> from my files, but that would not keep the correct date of the last
> change for unchanged files.
>
> According to the idea of putting somewhere else tools not under
> development, why not then having a complete reference to the whole
> Apertium project.
>
> I mean somewhere else outside github, there would be the last version
> of all the file of the apertium project.
> trunk, nursery, staging, incubator, languages, tools etc... would be
> sub-directories of apertium and would include language pairs, languages
> or tolls as other sub-directories.
>
> Updates from Github repository would be done at least once a day, but
> once every 6 hours or once an hour would be better.
>
> Except for this kind of update, this reference would be read only.
> Date of changes on files and on directories (for directories, the
> date would be the date of the last change of github repository, that
> would also be the date of the last changed file inside) would also
> be updated from github reference.
>
> With that, several things not done with github (like ls -l or a php
> equivalent) could be done there.
>
> > I estimate that any requests can be fulfilled by Shardul,
> > Tino or myself within 24 hours. Please let me know if there are any
> > repositories that you would like migrated.
>
> I would prefer first the problem of listing apertium branches and
> directories to be fixed, but a change should be done in the file
> desmnemo.c of apertium-c-formatters .
>
>
> > More information and rationale is available on the Wiki:
> > http://wiki.apertium.org/wiki/Migrating_tools_to_GitHub
>
> > Please let me know if you have any further concerns.
>
>
> > [image: Sushain Cherivirala]
> > *Sushain K. Cherivirala *
> > Stanford University, M.S. in Computer Science '19
> > Carnegie Mellon University, B.S. in Computer Science '18
> > (713) 992-4043  <713-992-4043>| www.skc.name
>
> --------------------------------
> Bernard Chardonneau (France)
> Phone : [33] 9 72 36 32 90
> GSM phone : [33] 7 69 46 16 31
>
> An alternative Apertium translation website :
> http://apertiumtrad.tuxfamily.org
>
> Multilingual websites for my free softwares :
> http://libremail.free.fr and http://libremail.tuxfamily.org
> http://cyloop.tuxfamily.org (mainly translated with Apertium)
>
> My general website (in french only)
> http://bech.free.fr
>
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Apertium-stuff mailing list
Apertium-stuff@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Reply via email to