http://freshcode.club/projects.rss has been showing duplicate "LimeSurvey 6.5.17 (build 240715)" since yesterday. :(
LimeSurvey entries are spamming the RSS feeds (and the atom feeds) with an entry being generated every 5 minutes even if there is no update to the code (so that entries end up as duplicates of previous entries). This makes the RSS feeds for the whole freshcode completely useless.
Would be great to find out the reasons for this and why LimeSurvey is the only affected project.
Is this originating from a problem on the side of the LimeSurvey developers? Namely, do they send status updates to freshcode even when there is no reason for them? Do the "spurious" status updates get somehow filtered for the website, but not for the RSS feeds? If the LimeSurvey project itself ends up as breaking freshcode, then it should be informed.
Unless the issue can be resolved, I strongly suggest getting rid of the RSS feeds altogether. Users would end up unsubscribing from them anyway.
I have added this project:
http://freshcode.club/submit/ursalang
I've tried various settings, but nothing seems to work. (I have been making pretty regular releases, so Freshcode has had plenty of opportunities to auto-update.)
The two releases of Ursa that appear are both manual submissions.
Freshcode version is up to date on the site: http://freshcode.club/projects/kdenlive (23.04.03)
But on repology it's showing 19.04.03 see https://repology.org/badge/version-for-repo/freshcode/kdenlive.svg
Repology says there is a discrepancy on freshcode's side of the equation. Please advise?
When selecting the language c# as suggested by the Add Tags Trove, it becomes c on save, which is obviously very wrong.
If the GitHub project uses calendar versioning, the auto-update mechanism will fail.
Since one of the last remaining OpenID providers folded, I decided to migrate to IndieAuth. And OpenID support will be removed sometime in the
near future. (There's just been a handful of remaining users.)
It's possible to keep using it though. There's a proxy/delegation option on https://indieauth.com/openid
For everone else, you can easily set up IndieAuth on your personal homepage, by chaining to your preferred social media site with:
<link rel=me href="http://twitter.com/yourname">
<link rel=me href="http://github.com/account">Or even set up an IndieAuth-only account.
I've prepared an authorization endpoint in the fossil repository.
<link rel=authorization_endpoint href='https://fossil.include-once.org/freshcode/ext/auth'>
Hey. Just wanted to drop a note that repology.org, a service which monitors and compares packages across a lot of repositories, has also support for freshcode.
It works both ways: information from freshcode is used to discover new software releases and thus packages which need updating, and information from repositories and upstream may be used to update freshcode. Here's list of projects outdated on freshcode, for instance.
however it's not complete as I'm only accumulating information from RSS, so it only contains packages update during latest 3-4 months. Is there a way to get full database?
Well, there have been a few (actually rather few) spammy submissions lately.
So, a big thanks to all who took the time to flag those recently!
Hmm. Seems that backslash character was eaten from "Changes". Even duplicating it did not worked.
( was on http://freshcode.club/projects/elmme25 )
Submit for do not tell which characters are reserved.
/ Kari Hurtta
Yes, all text fields are filtered somewhat rigorously. The according code is hidden somewhat in: http://fossil.include-once.org/freshcode/artifact/31d1e3102bc5ca98c759c7d489058e0a178534a2
Specifically it uses a regex:
/[^\w\d\s,._\-+?!;:\"\'\/‘´()*=#@]+/u
To filter non-regular-text. Which is too avoid too programm-y content. While sure, programmers is certainly a main audience; the changelogs are usually not that fine-grained or meant to be excessively technically.
"Terms and Conditions" say
Your entry is shareable under the CC-BY-SA license.
"entry" here refer to submitted text and not actual project or program ?
This is little ambiguous.
I think that it must refer only to submitted text, because other that first condition does not make sense:
It's open source / libre / Free software or pertains BSD/Linux.
( I consider submitting Elm Millennium Edition, was on http://freecode.com/projects/elmme
Technically license is "The Elm(tm) Mail System General Public License" aka http://www.elmme-mailer.org/NOTICE )
/ Kari Hurtta
Hi Kari. That's true; I just stumbled over it the other day myself. So I'm gonna reword that to "description / text".
The Elm license of course is also on-scope here. If you want you can flag it for moderator attention later, and I'll add e.g. "Elm GPL" as specific license marker to your project listing.
Also, thanks for the input and taking the time to resubmit your freecode project here!
I had problems with version numbers on freecode. One reason also was that they removed branches on site update.
How I give these versions so that release title have correct name ? This problem was also freecode (and sometimes staff edited they that result was unclear.)
Examples:
"Elm 2.4ME+ PL126 (25)" has filename elm-2.4ME+126.tar.gz
( on some library files that uses elmme.1.0.126 as version, but that is not visible program version.)
"Elm ME+ 2.5 PLalpha37" has filename elm-ME+2.5.alpha37.tar.gz
( In some deb files that uses elmme.1.1.37.{dist}, as internal package.version, but that is not visible program version. )
"Download URL" notes In particular for the download link one could apply the $version placeholder.
I guess that these branches is best register as different project ?
However I not be sure what use as "Project ID". It says
A short moniker which becomes your http://freshcode.club/projects/name. May contain letters, numbers, hyphen or underscore.
( Reason for these strange version names is that:
/ Kari Hurtta
Yeah, I'm sorry. But freshcode doesn't provide actual branch support either. It's rather uncommon for smaller projects to maintain two distinct version trees. So the submission page assumes there'll be a "2.5.001" and a distinct "2.4.001" version for instance.
You can either register the project twice (elm-24 as project id, and elm-25 for the newer branch). Or just create one project name elm, with a common project description, and later submit two releases. (This is what I'd currently recommend.)
The version numbers are mostly free-form. So you can in fact include the branch information there. For instance with a project id of elm you can submit version numbers of "ME+2.5.alpha37" and later "2.4ME+ PL125" as release.
Obviously the project listing will always only show the download/source URLs to the last entered release. If the version numbers differ in filenames, I'd advise just using plain URLs (without $version placeholder).
(While there is technically a way to support branches in our database [through the state: field], I haven't yet prepared anything to actually utilize/visualize it. There's however the releases.json autoupdate scheme to better control this branching method and release announcements.)
The version numbers are mostly free-form. So you can in fact include the branch information there. For instance with a project id of elm you can submit version numbers of "ME+2.5.alpha37" and later "2.4ME+ PL125" as release.
In other hand project "elm" is different. I do not claim to own it. And it needs different description.
My project elmme aka "Elm ME+" was http://freecode.com/projects/elmme
Project elm was http://freecode.com/projects/elm
There on project elm was versions "2.5.7" or latest seems to be named "2.5 PL8".
There is no possible give "printable release version" what is different than catenation "$project $version" ?
If not, then I will use full version string on project version string. For example
"Elm 2.4ME+ PL125 (25)" "Elm 2.4ME+ PL126 (25)" "Elm ME+ 2.5 PLalpha36" "Elm ME+ 2.5 PLalpha37"
Listing on front page will probably show as
Elm Millennium Edition Elm 2.4ME+ PL126 (25)
and
Elm Millennium Edition Elm ME+ 2.5 PLalpha37
( and download URL as http://www.elmme-mailer.org/elm-latest-src.html )
or if I make two projects (so that these have different description and links)
Elm Millennium Edition 2.4 Elm 2.4ME+ PL126 (25)
and
Elm Millennium Edition 2.5 Elm ME+ 2.5 PLalpha37
That way at least "Recent Releases" listing have same version names than what program shows.
Yes, this is little messy.
/ Kari Hurtta
You worry a little too much :) All the fields are basically treated as description fields, including the version number.
Of course you'd use elmme then as project ID for clarity. And if the project title is already "Elm Millenium Edition" you could probably leave out the redundant "Elm" prefix from the version field. If you maintain this yourself anyway, the exact serialization or the historic version scheme shift is not as important. As long as people know which version is current, it's all good.
While of course overlong version numbers should be avoided, the technical length limit is over 30 characters (so covers your case too). Version numbers are only cleaned up using a input::words() filter internally. This means A-z, 0-9, . and - or spaces ␣ etc. are retained. A few special characters like + as well. However parens ( ) or the # hash would be removed. So in your case the PL126 (25) suffix might better be written as PL126-25 right away.
However, don't worry too much about this. While the version field is somewhat of an exception (changing it usually publishes a new frontpage entry); you can otherwise edit project descriptions at any time. Unlike with freecode, you see the submitted text instantly, and can reedit as often as you like. Since you have a summary download page, the version number used here is not going to shroud this either way. -- If you feel it needs any cleanup later, just flag the entry; things can always be augmented or merged in the database.
(difficult to write when that form jumps up and down)
I ended to use "elmme25" as project id, "ELM ME+ 2.5" as title and "PLalpha37" as version.
/ Kari Hurtta
Hey, thanks again for your time. This scheme/project versioning approach looks indeed most user-friendly.
Sorry about the forum :/ - I'm working on the JS flippiness..
I added another with "elm24me" as project id, "ELM 2.4ME" as title, but without version.
Seem that still entry to "Recent Releases" was generated.
There are versions, but they are not very recent, so tried to avoid claiming that now is released new version.
I screwed something.
/ Kari Hurtta
That's intentional. Projects without release number will still be visible. Since freshcode.club is somewhat in its infancy, and reregistration from freecode.com projects is expected; even older project versions are welcome.
I've just set it to "hidden" on the frontpage now. It'll still show up in the search pages / tags etc.
I see. So I missed scope = "hidden". I added also "parent" project (project id "elm" to here). I think that state "historic" is adequate on here :-)
The last release is on 2005 and I think that the first release was 1986.
However parens ( ) or the # hash would be removed.
I forgot this ☹
/ Kari Hurtta
Thanks for the note. Fixed it back. (Your versioning scheme is somewhat of an exception.)
Your versioning scheme is somewhat of an exception
Yes. ☺
"Elm 2.4ME+" is (indirectly) based on "Elm 2.4" and that "(25)" was saying that "(security) patch PL24 -> PL25" of "Elm 2.4" was applied.
So that is somewhat historic tag.
/ Kari Hurtta
Thanks for the note. Fixed it back.
Forgot that edit, I changed wordpress -link and system generated new release. ☹
I changed wordpress -link and system generated new release. ☹
Now that "Elm 2.4ME+" shows correct release name "PL127 (25)", but now it show that this is entered "29 May 2016 13:43" when this release actually was entered at 2015 to freshcode.club.
/ Kari Hurtta
I've fixed it manually. I guess the original change was reverted after the server update a few weeks ago. Going to look into that..
Allowed version meta characters should be ~+()#/@: now.
Sorry, everyone! We had a server outage last week. And it took some time to migrate everything.
Also not to worry, with freshfoss / fc-mirror / etc, there are enough distributed freshmeat successors meanwhile..
Hi,
I just recognized that the mirror functionality I have installed at http://freshcode-mirror.opensourcestore.org/ is broken.
Can I update the mirror script files from current release?
Thanks, Lothar
PHP Warning: Invalid argument supplied for foreach() in /srv/www/htdocs/cron.daily/xfer.php on line 27 PHP Warning: file_get_contents(http://freshcode.club/img/screenshot/.jpeg): failed to open stream: HTTP request failed! HTTP/1.1 404 Not Found in /srv/www/htdocs/cron.daily/mirror-screenshot.php on line 67 PHP Warning: file_get_contents(http://freshcode.club/img/screenshot/cmarkc.jpeg): failed to open stream: HTTP request failed! HTTP/1.1 404 Not Found in /srv/www/htdocs/cron.daily/mirror-screenshot.php on line 67 PHP Notice: Use of undefined constant CLOUDFLARE_TKN - assumed 'CLOUDFLARE_TKN' in /srv/www/htdocs/cron.daily/header_stats.php on line 24 PHP Notice: Use of undefined constant CLOUDFLARE_EMAIL - assumed 'CLOUDFLARE_EMAIL' in /srv/www/htdocs/cron.daily/header_stats.php on line 25
If the updated entries still appear, I don't think the rest of the syncing fails.
There's just a few missing images. Your screenshot update script just polls for non-existant projects. cmarkc was renamed to cmark.c IIRC. And the empty .jpeg is probably a missing project name somewhere.
Maybe the mirror code doesn't use the recent RewriteRule set. I've introduced faux file extensions for project names after branching that off. So you might want to update some core files like .htaccess and aux.php indeed.
The CLOUDFLARE_* stuff is only used for the website stats. You should disable that cron part perhaps.
Hey, let's just set up a screenshot archive instead. That's probably easier than fetching all images one after another.
I'll setup a cron script to generate http://freshcode.club/img/screenshot/ALL.txz instead.
I plan to move my code, but do not like to loose my history, what seems difficult for CVS. That's why I am still using sf :-(
Some point's are also my points, thus I agree on the overall lost of trust: https://sourceforge.net/blog/sourceforge-acquisition-and-future-plans/
Regards,
Lothar
Just a quick note. There's going to be a bit of planned downtime on Monday, 2nd November.
The VPS is being updated or something something software update.
Hi, first a big "thank you" to the freshcode developers team for this excellent service! It looks like I mistyped or confused some other way my password when I submitted my project, and now I do not seem to be able to add a new release, which puts me in a somewhat awkward situation. To tell the truth, I have been thinking about submitting the project a second time under a slightly different name (and this time without lock), however before clutching at this no less awkward straw I felt like asking if I missed something obvious. Since I might not be the only one to suffer a sudden fit of Alzheimer's in the summer's heat once in a while, I thought I might put the question here as well. Sorry for bothering you with this, and a nice week-end to all of you!
Hi Michael. Thanks for the note. (And your project of course!:)
There isn't yet any password reset function. Mostly because there are (purposefully) no real user accounts and email addresses in the database.
(It was originally just meant for OpenID. Which is also the reason it's a rather confusing UI for setting a password in the first place. I'm going to add a new dialog or something to simplify both.)
Otherwise, simply send a mail to info@freshcode.club. Best from the same address your GitHub or SourceForge project is registered to / or your project@domain even. And I can just remove the password/hash for example.
Btw, it's not an issue if there was a duplicate project entry. Just drop a mail, or flag for moderator attention, and it easily could be merged back together. (A handful of projects use multiple project names for branching. Also perfectly fine.)
It should be for tweets around open source. It's not only for release information, but related to all around oss.
Feel free to follow :-)
https://twitter.com/ossretweet
Lothar
[finished: 2015-05-25]
Just a minor tweak to the polling mechanism, and I uncovered a "little" backlog:
(A few of those are minor/hidden releases though. And some are very overdue.)
Now usually frontpage time for each project should at least be 2-3 hours. But this weekend it might look a bit like rush hour. Sorry! -- So perhaps hold back fresh project announcements till Sunday.
This isn't completely done. But starting from now, it's possible to use a plain password to lock up project entries. OpenID isn't overly popular, so we're going back to basics here.
Please note that registering/updating a project is a "two"-step process:
That'll fixate a weird text string in the lock field. Which is quite simply your personal password hash.
(More technial: you can have multiple passwords. The "lock" field is a comma-separated list. It can hold multiple crypt/bcrypt hashes, and alternatively OpenID references.)
So, I've been deliberating this for a while.
What our database still lacks is an alternative short project summary. Besides the title and the full description field, there should be a shortened abstract for each project.
This just came up (again, I think) when discussing the JSON submission API. The old freecode API exposed a "oneliner" field, but actually didn't use it much in the UI.
Yet for freshcode there are various places where it totally makes sense:
Now what I'm not quite sure of is how to nicely fit it into the frontpage releases. The olden Freecode-style layout doesn't really help there. → If you have a suggestion or mockup on how the frontpage should look with a short extra summary for each project, please share. T!
Anyway, I'm going to start updating project records - before we pass the 1K milestone...
This forum (it's threaded, not a board; also uses MarkDown CommonMark) is intended to:
And obviously FLOSS developers are also invited to e.g.
Or whatever else comes up your mind.
Currently project names (not the titles) are constrained to foo-bar or baz_2000 style identifiers.
Would it make sense to permit e.g. http://freshcode.club/projects/foobar.py names as well?
py-imlib" and "perl-imlib".Edit: done. All projects can now use basename.lng as project id.
Since GitHub only provides a somewhat stale trending repository list, we now have a project release overview (shortened version in the frontpage feeds).
http://freshcode.club/github-releases lists tagged announcents, and somewhat eases uncovering well-maintained projects on GitHub. The filters still need some improvement, obviously.
Just like the GitHub filter, there's now also a /launchpad-releases. Interestingly more projects actually utilize the changelog feature there.
It's slightly less current, as the polling takes some time. (The LaunchPad API is quite cool, but I don't want to overtax it.)
While “fcc” is already somewhat known amongst us programmers, it doesn't quite match up to the former popularity of freecode/freshmeat with end users.
To get a little more publicity for this aggregator and all our projects, let's start 2015 with a wee advert campaign. This is what it currently looks like:
The technical headline is intentional. And it's restricted to English-language searches for now; also set to show up on AdSense blogs/etc only rarely.
← (In case you missed the logo rotation ;)
I tried to slashdot a small hint, but that was declined. So simply add a hint for each visitor at the start page to kindly help in spread the word.
Simply move the social bookmarking at the top...
Of course having the social bookmarks on top would help. Still considering if it can be integrated with a nicer layout somehow. I'd even want to expand the logo with the „Collab-Map“ I've been pondering for a while. It would be less obtrusive, prettier and more engaging there. (The current pseudo-icons look somewhat odd anyway.)
As far as social links are concerned, it's not that bad. Social.ajax.php sayeth:
And Google already counts 6,519 total backlinks.
I'm also gonna try a Slashdot story when we reach 1000 projects. (I've been somewhat lazy with adding new entries myself last month). But as far as project homepage backlinks are concerned, I'm going to assemble an Excel/Calc sheet to prepare some manual mails to developers/maintainers. (Yes, that'll take a while. No emails in the database, and it's gonna stay that way; so needs to be done the long-winded way.)
Btw, have you seen the new HTTP 500 error page yet ;? - It also lists OpenSourceStore (I'm going to add some titles= or make it more descriptive still..)
There is space that result of "Browse" button is shown on several column. Now result is shown as one column which uses quite small part of screen. Depending of client there is space for several columns. On desktop at least for 2-4 columns.
And your layout does not seem anyway work very well on small sizes.
Well that layout seems not work with my small laptop.
screen #0: dimensions: 1024x600 pixels (271x159 millimeters) resolution: 96x96 dots per inch depths (7): 24, 1, 4, 8, 15, 16, 32
Ok, it works if I press F11 (full screen mode). At least it is close to usable.
/ Kari Hurtta
Hi again. I've been meaning to rebase the layout on e.g. Bootstrap (assuming that's not the origin of Twitters´ sluggishness). The current layout suffers from the wrong mix of fluidness and fixed boxes. It doesn't work on mobile, and with less than 1000px the sidebar floats into the #main section.
However, as the stylesheet is a little close to freecodes graveyard theme; I'm still pondering if a complete redesign would make more sense. There's a few more technical issues in the freshcode backends, which is why I've been putting this off. (It's number #4 #3 #2 on my ToDo list however.)
I think that it maybe little better if text in project description and changes flow around project picture.
Now picture just reduces width of description and changes -box.
I've already adapted the screenshot generation for slightly larger and nicer previews. And this would of course impact the description+changelog flow text further. (It often looks okay still with 4-5 lines of description text).
So a future layout would probably put the screenshots besides the project descriptions. Possibly along with the tags. → I'm imagining small tag clouds besides projects, which would also help finding projects per language quicker. → And lastly I'd thought about two-row project lists on the frontpage. There's enough whitespace to reuse for that. This would put more project listings above the page fold and thus in the spotlight.
Hi,
thinking about feedback, how much the site or a project is clicked, is it possible to generate a diagram for the clicks per month/week?
Giving a small hint about the reach of this site would be a good information.
Thanks, Lothar
There's no stat counter or anything yet. All I can offer are some access log summaries.
This weeks (since monday) access.log shows 3000 visitors to the frontpage, and 1200 on per-/projects/ pages after filtering out search engine spiders. The RSS feeds seem to be quite popular, btw.
What we can't tell is how many people directly click to project homepages though. (Which I'd hope to become the main attraction really). Unlike freshmeat there's no URL-tokened click counter for that, albeit a ping= might be feasible and acceptable.
But true, more concrete infos make sense. Maybe I'll just set up a proper log analyzer. Though it's probably senseful to have something more customized later to differentiate between active views on current releases and those that aren't on the frontpage.
I think, that starting with a simple counter per week in the main page will be a good starting point. Adding another field in the project entry to count page hits may be a simple next step to simply display that counter for the project.
Is the meta page currently the only feedback mechanism, or do you know any other place that has emerged as a discussion platform?
My forum is yet not as successful as I am like it to be.
Lothar
I mostly thought of meta to be the local feature request survey, because bug trackers are commonly less enticing.
As for OpenSourceStore, I originally envisioned it to progress into a neat e.g. online-shop styled Freshmeat alternative - just as the domain already sounds a bit like.
Now that FCc seems almost ready for a beta stage, would you be interested in setting up a customized copy/frontend and database mirror? Your existing forum could prominently list the existing discussions in a feed pane (a bit prettier than the article/software links here of course), which might draw more discussion/attention. Or else maybe convert the entries and turn the forum altogether into a user-contributable blog w/ software reviews?
I thougt about this, but first I tried to get up some constructive discussion. Maybe a mirror + discussion + some more users with approval permissions are good. I had a look at your feed mechanism and it seems there is no one with a complete copy, may be due to performance reasons. May be there a REST API with paging support and or syndication?
Actually I do more like fighting phpbb spammers :-)
And yes, the domain sounds like a shop. Each oss project needs some money - maybe paying for the website, providing little more predictable support and the like. But this is a concept still in my brain :-)
A big company that has a business concept around oss may not need such a site, thus it is thought for individual oss developers seeking for support to pay for or providing support by participaing there.
There is another site doing similiar stuff: http://www.myopensourcestore.com/
So while I am thinking about a concept others started up and doing it before me. I'll investigate some work figuring out what they do and what can be done better.
Lothar
The current traffic is comperatively negligible, as the templates are rather slim, and CloudFlare optimizes and caches away the rest. Only the local GitHub/Archive API polling consumes a few more GB, but that's about it.
As for mirroring, there is already /feed/xfer.json?num=500. That'll stay even if the other feeds become static, no performance issue really. Adding a sliding parameter is no effort, but not needed yet. And btw, that JSON feed is almost a raw database dump; thus can be consumed by a trivial import cron job. // Setting up a phpBB or $_SESSION bridge (as alternative to using OpenID) to the freshcode 0.7 implementation should be likewise simple.
Can you tell me what steps are required for setting up a mirror?
I assume to have to keep my own copy of database to store fed data I got from you, but then how to redirect update and insert requests to the master?
If I do not have a database, how to I allow adding releases?
Or should this be done at the master site only for simplicity?
In all cases I assume to modify at least configuration, but maybe code.
It's not yet thoroughly prepared for separate setups, but should be easy to customize. I'm going to branch a mirroring tree (including a robust xfer-sync cron job). As dependency PHP 5.4 / HHVM 3.0 is required at least; wkhtml+imagick for screenshots, and SQLite obviously (it still needs a database, even if project submissions were turned off).
Theme-customization might be a bit involving, as the stylesheet got quite broad, but the templates/* shouldn't be. Using it initially as read-only frontend without any management overhead is comperatively simple: the [submit] and [update]/[flag] links can just point back as absolute URLs to FCc.
Hi,
I have some problems getting freshcode installed on my developer box. Mainly due to probably misconfigured / unconfigured phar feature in php.
A setup instruction should be added into the wiki at least, so that it will become a nobrainer for dumbs like me :-)
I'll try it by reading the phar docs, but you may be faster telling how.
Lothar
I've prepared a freshcode-mirror.tar.gz just in case (just fixed the anon-fossil zip download permissions, sry).
Should be easier to set up (empty database included), and just requires PHP 5.4 or something.
I encountered the second time a small error when I post a reply. When I have done my reply and follow, it seems I have overridden the post I am replying to. Only a page refresh shows me I am wrong and all is ok.
I couldn't yet be bothered to fix the DOM insertion. (Basically the Reply/Edit/Submit hooks use the same API call, but don't differentiate which ul/li node to attach results to). Meta only uses AJAX to deter common spambots (not foolproof, but a sufficient diversion).
Took a while, as I couldn't really use the local webserver log as counter. The CloudFlare (used as CDN proxy here) stats were more realistic:
Albeit for the new summary in the header (got commited a bit too early), the numbers are only approximated conservatively. The for-free CloudFlare API just gives out the last 24h hours. But it's reasonable enough for a quick and unpretentious overview; project and release numbers seem more interesting anyway.
Unlike freshmeat there's no URL-tokened click counter for that, albeit a ping= might be feasible and acceptable.
Hmm. Seems that ping= exists does not exists on HTML 5 (as W3C Recommendation 28 October 2014).
OK, it exists on whatwg's HTML.
OOPS.
Seems that ping= exists does not exists on HTML 5 (as W3C Recommendation 28 October 2014).
Should be: Seems that ping= does not exists on HTML 5.
Mozilla documents it https://developer.mozilla.org/en-US/docs/Web/HTML/Element/a as available. So it's probably just the W3C documents which are out of date.
Hi all,
I have setup a - probably first - mirror for the main site with some help of Mario.
Performance is an issue I like to measure with your help regarding the usage of the CoralCDN caching.
On my home network I experience very slow performance and on my tablet with UMTS it is even worse.
Can anyone report on this issue?
Here is the link: http://freshcode-mirror.opensourcestore.org
Thanks, Lothar
"Can anyone report on this issue?
Here is the link: http://freshcode-mirror.opensourcestore.org"
I just got
Server not found
Firefox can't find the server at freshcode-mirror.opensourcestore.org.nyud.net.
( Looks like it resolves to http.l2.l1.l0.nyucd.net, but http.l2.l1.l0.nyucd.net does not found.
When I test from different places than I got some IP addresses for http.l2.l1.l0.nyucd.net, but also error code NXDOMAIN is returned. )
Some tests
{ ... I removed this post because I do not understand markup ... }
As you see DNS gives errors and when IP has got, then there is no response.
Beautiful. ☻
Not exactly sure how it works. But CoralCDN turns out to be pretty slow for me as well. I've actually used it a few years back for another project where it was working ok. Why it's not living up anymore, no idea. Might just be provider-dependent, at least in my case.
There may be issues with my private box, that is running the site. The upstream speed is about one MBit I think.
I have added mod_redirect stuff to the server but there is still missing a mod_deflate feature (aka compress). I don't know if this is a solution regarding coralcdn, but it will in general help improve the speed.
The main issue with a move is simply to move from development to production and I have a webspace to be used for it :-)
Another issues are found by using google web page speed tool, so I'll try to fix stuff and report on stuff to be merged back.
I'll start with that in the next year :-)
Seems that "Search" button does not found project, if project id is given as search term.
At least "elmme25" and "elm24me" does not work as search term and they are project IDs.
"elm" works, but that is on project title.
/ Kari Hurtta
And of course it works if I use full title as search term, ie. "Elm ME+ 2.5" and "Elm 2.4ME+".
We didn't have many cases of that yet, so searching by project ID wasn't accomodated for. However it's easy to add. I'll just add name to the list of database fields to search in.
More concretely, this features isn't yet based on an actual search function (SQlite FTS et al). But often works okayish enough.
Update: Committed an initial patch for just matching the plain name field as well.
Because the submit form documents just the basics, here is a longer explanation of what each field does.
→ Project ID:
Becomes your project URL name on freshcode. You can use letters a-z, numbers 0-9, hyphen - and _ underscore. It may additionally end in a language tag .py or .cpp for example.
It's supposed to be your projects basename in the URL http://freshcode.club/projects/PROJ-NAME.
Spaces and inacceptable characters will be stripped on submission.
→ Title: Is the full name of your project. While you can add a little elaboration, try not to use it as description field already. (We'll add a separate "oneline" or "summary" later..)
→ Homepage: Should be your projects homepage URL, or github or sourceforge reference.
→ Description: Explain what your project is about. List all the features. It doesn't have to be a lengthy text, but should round-up what advantages your implementation has over similar apps. You can even add future goals, or anything that may capture user interest.
→ License:
Pick one of the existing open source project licenses.
If there's no exact match, chose "Other" initially, and flag for moderator attention. Custom license names can be added still. Distributions may often pick "Mixed" instead; if the project is covered under different individual licenses.
→ Tags:
Should list a few categories to describe your project. Add the implementation language, core features, a target audience descriptor (end-user, developers, administrator etc.) if you want.
Note that tags should be made up of letters primarily, or - hyphen-separated categories. You can list multiple tags by separating them with spaces and/or commas. Such as c++, python, graphic, editor, image, end-user.
You can also pick a few general ones from the Topic/Trove box.
→ Image:
If you have a screenshot or logo for your project on your project homepage, just add an URL to it here. This will be used as preview.
If you leave this field empty, don't worry, a homepage screenshot will be generated then instead.
→ Version:
Your current project version. (Leave empty if there isn't one yet).
Version numbers may be made up of numbers, dots, hyphes, underscores, and -xyz suffixes. This field however also accepts textual version descriptors.
Note that you'll generate a frontpage entry whenever you add or change the version field.
→ State: Broadly categorize stability or finishedness of your project.
→ Scope: New releases can be minor or major feature changes or bugfixes. What qualifies as minor or major may be interpreted differently by each project. (For example the Linux kernel adding 100 new drivers might be minor to them, but a small CLI tool changing some behaviour might amount to a major feature change).
→ Changes:
Is a summary of what new features or changes and bugfixes your newest release adds. Write this end-user readable, in a list style, but avoid the * decorators if possible.
→ Download URL:
Is either a link to a zip/tar, or a download landing page.
If your zip/tar file releases consistently contain the current version number of your project, then you may wish to utilize the $version placeholder. - So for future updates you don't have to update this download URL again.
→ Other URLs:
You can add further project links. This field accepts a line-wise or comma-separated list of Link-Name = http://example.org/ fields.
You often may want to add GitHub or SourceForge URLs here, or a link to the manual/documentation, or version control system. You can just as well list DEB or RPM or APT-PPA download links, or release notes, a user forum, or a mailing-list= entry.
See the Autoupdate Howto..
→ Submitter:
Just give your name here, or a nickname of yours. You may additionally (comma-separated) list a registered gravatar@example.com address, or use your username@github, username@sourceforge or username@launchpad. If so, a user icon will be presented later.
→ Lock:
If you've logged in beforehand, then just click on the greenly-dotted lock field. This will copy your exact OpenID id into this field. Which in turn will prevent editing of your project submission by anyone else. (Except editors/moderators on freshcode.)
→ Terms and Conditions: This section just requires to check two boxes. It's mainly an obfuscation for spambots currently. (No need for Captchas or really requiring OpenID yet.)
One of the core features on FCc are automatic release updates. (Can't stress this enough ;). The project fields however look more complicated than it actually is.
If you've got a CHANGELOG or NEWS file (in your repository or homepage), than just enter that under Autoupdate URL. And set the method to [Changelog].
The common format of NEWS files is to have the newest version on top. Each version entry gets a header and a list of changes:
1.2.0 (2014-12-31)
------------------
* Added new feature xyz.
* Fixed bug with --foobar.And that's it.
But take care not to include unreleased versions (or tag them with (UNRELEASED) instead of a date.)
If you've got a non-standard format for version headers, then you can often easily devise a regex for that, e.g.:
header = /^Project Name released ([\d.]+)/mWhere [\d.]+ is the most common placeholder to match numeric-dotted version strings.
All projects can utilize a Changelog or a homepage News section to keep listings automatically updated. Previously the autoupdate process ran nightly at 3:15 GMT. But it dumped all found version and project changes at once.
This reduced frontpage publicity on busy days. So I've recently adapted this process. New releases will now be spooled, and instead get published with at least 2 hours spacing each.
While this may delay announcements half a day further, it prevents release lumping. And it IMO gives each project release a fairer time slice topmost on the main index.
(The overall autoupdate process still needs finetuning obviously. The long-term plan is to have it run more than once per day. It just incurs a few database lockups still, and I didn't want to mercilessly poll project homepages yet. Btw, the autoupdate code and modules are not yet in the repository because of a few other woes.)
We now have an implicit project announcement channel on Twitter:
https://twitter.com/versiontracker
All our project releases will be squeezed into 140 characters (including a direct project homepage link).
I've been planning this for a while, just went forward with a preliminary version for now. One of the inherent goals of freshcode.club is to relieve project maintainers of a few tedious tasks (hence e.g. automatic updates per Changelog). And extra Twitter announcements are one of them (not many projects do, due to the setup effort -- or at least I for one never bothered).
While YourProject 1.2.3 released. http://example.com/ #cpp #music #recording is not as thorough as the complete announcements on our frontpage, it's sufficient for generating a little interest for twitsumers/subscribers there. (Coincidentally it raises the social networking ★ star count, as depicted in project descriptions.)
The twitter bridge cronjob is currently using twidge for simplicity. The supported URL shorteners do not allow for image screenshots to be included; Twitter seemingly requires the online t.co for that.
twurl, so at least the mini-screenshots are now embedded.Likewise doesn't it generate and include @project references yet, for those packages that have a Twitter= URL entry. (It's just a handful in our project database however.)
Please add two license tags regarding the MPL. MPL 1 was created by Netscape and is not compatible with GPL, etc. MPL 2 is a simple Copyleft license that was explicitely designed by the Mozilla foundation to be compatible with GPL, LGPL, AGPL. That's a major difference for code reuse and thus important criteria for categorizing FLOSS projects. So please allow the MPL dropdown to differentiate between the two versions.
In this very case I'd retain the "MPL" moniker for the current version, and add a historic "MPL 1.x" entry somewhere in the list below.
PS: Also if anybody wants to have a specialty license listed, just pick "Other" initially, and flag the entry for moderator attention. Custom license names can be added individually per project.
Thank you for creating an maintaining this site!
It would be nice if there was RSS feeds available for each project and for the entire site as well.
Cheers Johan
Hi, and thanks. There are a few RSS feeds already, they're just a little hidden.
I'm still working on generating static feed files, which can be better integrated with project listings then. (It's only going to be RSS however, Atom support will be dropped.)
Currently freshcode.club is stylistically close to the late freecode. It's however kinda obvious that projects weren't presented prominently / pivotal enough.
One minor difference is that homepage links are slightly more accessible already. Not just the small icons do, but the preview images directly point to project homepages.
The overall frontpage design IMO however still weights the directory too high.
ProjectXy 1.2.3 title bar links need a rethought.If you want to share some ideas (or better even: a mockup), please tune in.
Freecode.com was long established as central service to announce open source software releases. In June it suddenly shut down. Larger projects still can rely on blogs and news sites to circularize progress and new versions.
GitHub and Co. meanwhile do an excellent job at connecting developers. Distribution package managers cover software searches for end users even better. (Not just because the packaging systems grew up, but because of our relentless package maintainers. Thanks guys!)
Smaller and new open source projects however depend on distinct publication channels for recognization and attracting contributors, to avoid getting drowned amidst millions of other GitHub repositories. This is where freshcode.club is meant to pick up again.
It's obviously anything but finished, but there are already a few significant changes in contrast to the old freecode:
releases.json for publishing.There's quite a few areas that need further improvement. Such as the tagging scheme (both the UI as well as the Trove classifiers).
While it's a bit too early for featuritis, having a ToDo survey is not. So what would you like to see?