specto-devel Mailing List for Specto
Brought to you by:
kiddokiddo
Menu
▾
▴
specto-devel — this is notifying new ideas / technical matters to the devs
You can subscribe to this list here.
| 2005 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
(1) |
Sep
|
Oct
|
Nov
(5) |
Dec
(8) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2006 |
Jan
(20) |
Feb
(8) |
Mar
(8) |
Apr
|
May
|
Jun
|
Jul
|
Aug
(9) |
Sep
|
Oct
(2) |
Nov
(1) |
Dec
|
|
From: Kiddo <nek...@gm...> - 2006-11-12 23:13:11
|
Attention everyone! Changes ahead! First, what happened today? Thomas McColgan & me worked on getting Specto to run locally. That means you don't need to install it. This actually BREAKS setup.py and all the installation stuff. It needs to be fixed, if anyone wants to do it in their spare time, it would be good (especially for distro packagers). Second, we changed the directory structure a tiny bit. It is now cleaner. Removed some useless files such as "NEWS" and "README" which were empty anyway. Then, we moved a lot of stuff into a data/ directory, the source code in src/ and src/specto, and the translation files in po/. To run specto, you just need to use launch.sh. As of right now, the pygettext tool to extract translatatable strings from the code seems broken, for no decent reason at all it seems. If anyone wants to find out the problem and fix it, it would be much appreciated. Finally, as I could not get any outbursts of rage and indignation, and everyone seemed to consider resistance futile, I moved specto to Google Code today. What this means: - Specto's Sourceforge CVS is DEAD. Don't use it. Use the spanking new SVN repo, it works the same (but faster) and I updated the instructions on the wiki. - The mailing list will also be deprecated (this should be my last post?) in favor of google groups: http://groups.google.com/group/specto and http://groups.google.com/group/specto-commits. If all went well, I already subscribed you on those if you were subscribed on the sourceforge mailing lists. Please test :) - We don't use launchpad for bugs and features (specifications). Everything is tracked inside Google Code's issue tracker. I migrated the bugs and feature requests there, feel free to change them, claim the tasks, etc. - You need a gmail account for developer access. I can send hundreds of invitations, so please ask if you need one, I will be glade to provide it. If you don't like using gmail, well you can always tell it to forward all your mail, or use it through POP with a mail client such as Evolution. - You should remove your CVS directory and grab the SVN trunk from scratch. It will be a LOT cleaner, because of the changes we made to the directory structure. This is a copy of the post I will be sending to Google Groups. Jeff |
|
From: Kiddo <nek...@gm...> - 2006-10-26 01:36:06
|
Well, Since it seems everything was pretty much in order, I put some release announcements for 0.1 and a few packages in the file tracker (additional packages for more distros are welcome). This is the first step towards telling the world of our existence :) of course the development does not stop here. There's plenty of stuff to hack on, and now that we have a "fixed" release done, we can continue fixing problems without worrying so much. I put the translations template on rosetta, just for the curiosity of seeing if it would work. It is currently in the import queue. https://launchpad.net/rosetta/imports?target=products&status=all&type=pot If it works, it should appear in https://launchpad.net/products/specto/0.2/+translations Now, I'd like to ask for your opinions on the idea of moving from SF CVS to google SVN. I did the test, and it does need a gmail account. Now, obtaining one is not a problem: I can give hundreds of them to anyone who needs one (and I think it is now possible to create a gmail account without an invitation). The only problem I see is that... well you need to use gmail. I personally like gmail, but maybe not all do. What could be done to work around it is have gmail forward everything to your address of choice. Or use it through a mail client with POP3. Or use Specto to be notified only when you have specto-related mail in your account ;). In any case, we have people locked out of our CVS because they cannot be added as members of the project, which annoys me deeply. The current main things I would like to start fixing after that release is threading support (which we badly need) and full translations (glade files are not translated). Some fun ahead! Jeff |
|
From: Kiddo <nek...@gm...> - 2006-10-25 06:08:20
|
Hello folks! As you may have guessed with the slew of commit spam mails tonight, we now have (partial) translation support :) as bugs seem to be pretty much under control, I call for a nice 0.1 release ...tomorrow (well today, it's 2 AM here). This means that if you want Specto translated for that first release (nothing prevents us from making releases often though), you have to grab a .pot file and start translating the few strings we have :) If you intend to do so, please send a reply so I can be aware of your plans and "officially" release at the right time. If all goes well, we could have an Ubuntu edgy eft package (from Raof), an Arch package (from Willysilly) and maybe a gentoo ebuild too. I basically plan to put tarballs/debs/ebuilds/arches on the sourceforge file tracker, post on ubuntu forums and update the website for the announcement. Any other ideas? Suggest away! Now I'm pretty darn certain I will, as always, have forgotten to mention something and will realize that only when I click that Send button :) Jeff / Kiddo |
|
From: Kiddo <nek...@gm...> - 2006-08-30 18:29:58
|
Okay, I finally gave it a small try today, and I'm pretty happy with it :)
great job Wout ! What needs to be done now is merging this into specto with
the user interface and all that. Also, the logic that you and I discussed a
few days ago (about determining if the file was moved/renamed) should be
implemented at the same time.
Except that it tries to access the ~/.specto/cache directory even if it does
not exist, I have no bugs to report yet.
On 8/28/06, wout clymans <wou...@sc...> wrote:
>
> can somebody please try this and tell me what you think or file some bugs?
>
> usage is very easy:
> from watch_file_test import file_watch
> w = file_watch("/path/to/file/or/dir")
> w.update()
>
> and now remove, modify and create files and call the update() function
> again!
>
> thank you for testing...
>
> Wout
>
|
|
From: wout c. <wou...@sc...> - 2006-08-28 17:34:17
|
can somebody please try this and tell me what you think or file some bugs?
usage is very easy:
from watch_file_test import file_watch
w = file_watch("/path/to/file/or/dir")
w.update()
and now remove, modify and create files and call the update() function
again!
thank you for testing...
Wout
|
|
From: Kiddo <nek...@gm...> - 2006-08-18 06:02:05
|
Hello folks, Chris "RAOF" Halse Rogers kindly offered his help for maintaining an ubuntu package for specto in his repositories (see "installing" in the wiki). He suggested, however, that we start keeping the ChangeLog file up to date. That just means that before committing something, we should write down the interesting parts into ChangeLog, if they are of importance (new feature, bug fix, anything that we *finish*). So.. yep. |
|
From: Kiddo <nek...@gm...> - 2006-08-15 03:25:47
|
Hello, I would like to have your input on a decision to kill the instant messaging watch after I was talking a bit with William Rea about a Maemo port and all. My question is, does anyone not want to remove the planned feature of being able to watch instant messaging? My concerns are: * that feature does not actually exist * it would be tough to implement, and maybe not into Specto's scope of features? * I'm not really sure it would be useful, instant messaging systems pretty much notify you in every way they can * I don't actually remember why I wrote that down as a feature to begin with :| So, does anyone have comments, objections, cries of rage? If I receive no negative replies, I will assume you guys are okay with it, and will create a spec to remove that feature (from the documentation and the user interface, and also in the code backend), and hopefully do it myself easily ;-) |
|
From: Kiddo <nek...@gm...> - 2006-08-11 15:52:34
|
So far, seems good for me, but I'd like other's input on this too. The only
suggestion I have though is that the log file should be saved in ~/.specto.
The reason I think this is that specto is a user-level application, it is to
be used by multiple users on a single computer (remember our world
domination purposes?). I think (but I may be wrong there, it's just a guess)
that /var/log and such are places for big "daemon" applications such as
apache (that cannot be controled by anyone else than root or its
equivalent?) to store the logs.
In the case of specto, different users may have different problems to
report, and so I think the logs should be kept in their home directory (oh
yeah, please make the permissions the same as watches.list... actually, the
whole ~/.specto folder should be RWX permissions only for the owner,
security-speaking). Because one user might have problems, whereas another
one would not, on the same computer.
Thoughts?
On 8/11/06, wout clymans <wou...@sc...> wrote:
>
> Hello,
>
> I started working on a logging system for specto and now i need your
> approval to implement it in specto.
> The idea is very simple:
> *from logger import logger
> w = logger()
> w.log("watch updated", "info", "specto.watch")
>
> *so the syntax is: w.log(message, level, sender)
>
> there are 5 levels:
> debug, info, warning, error and critical.
>
> debug is only shown in the terminal, everything else is shown in the
> terminal and saved in a log file ( /tmp/specto.log )
>
> the format from the terminal is this:
> specto.watch INFO watch updated
>
> the log file looks like this:
> Fri, 11 Aug 2006 13:44:24 specto.watch INFO watch updated
>
> tell me if you want the format from the log file or terminal to be
> different, or if all levels should be saved to the log file or whatever,
> just let me know!
>
> questions i have:
> where should the log file be saved? (/var/log/specto, ~/.specto/, ...)
> should we keep the log class as a different file or are we going to merge
> it with another file?
>
> Wout
>
|
|
From: wout c. <wou...@sc...> - 2006-08-11 14:48:31
|
Hello,
I started working on a logging system for specto and now i need your
approval to implement it in specto.
The idea is very simple:
/from logger import logger
w = logger()
w.log("watch updated", "info", "specto.watch")
/so the syntax is: w.log(message, level, sender)
there are 5 levels:
debug, info, warning, error and critical.
debug is only shown in the terminal, everything else is shown in the
terminal and saved in a log file ( /tmp/specto.log )
the format from the terminal is this:
specto.watch INFO watch updated
the log file looks like this:
Fri, 11 Aug 2006 13:44:24 specto.watch INFO watch updated
tell me if you want the format from the log file or terminal to be
different, or if all levels should be saved to the log file or whatever,
just let me know!
questions i have:
where should the log file be saved? (/var/log/specto, ~/.specto/, ...)
should we keep the log class as a different file or are we going to
merge it with another file?
Wout
|
|
From: Kiddo <nek...@gm...> - 2006-08-07 18:22:35
|
There is still the problem of the google accounts. I do wonder, as nobody else replied yet, if this could be a problem. There may be people who don't want yet-another-account? Though to me it would sound a bit surprising to not want a google account ;) it may be perceived as a "devs are forced to go with a big company that owns a lot of services", though I don't think it's much different than having to use a sourceforge account. It would have the added advantage that everybody could easily communicate with jabber, for those who don't already have a jabber account. Although I _assume_ that google's svn repos should be fast and responsive, I am unsure we can use the same easy mechanism such as ssh keys that allowed passwordless logins and commits on sourceforge. Any idea if google's subversion does that Pascal? (http://subversion.tigris.org/faq.html#ssh-auth-cache). I have not found anything about that in google's FAQ. On 8/7/06, Pascal Potvin <pas...@mo...> wrote: > Flames are not for me... > Anyway, I always thought of google as a good place for handling my > mails, my jabber things, my own personal life. I never estimated > google as a place for working on open source projects. > > Still, I think the interface that google is giving us can be seen as > good. We can (yes we can) take the risk of putting everything there. > If somethings goes wrong, we'll just have to merge the new code on sf. > > My 2cents :P > > Pascal Potvin > |
|
From: Pascal P. <pas...@mo...> - 2006-08-07 06:57:24
|
Flames are not for me... Anyway, I always thought of google as a good place for handling my mails, my jabber things, my own personal life. I never estimated google as a place for working on open source projects. Still, I think the interface that google is giving us can be seen as good. We can (yes we can) take the risk of putting everything there. If somethings goes wrong, we'll just have to merge the new code on sf. My 2cents :P Pascal Potvin |
|
From: Kiddo <nek...@gm...> - 2006-08-02 22:13:39
|
Hello fellows ! Long time this mailing list has not been used hm? ;) As you may already be aware, Google launched a project code hosting service recently. I am personally beginning to get interested in this opportunity, because of the reasons I will list below. Anyone is welcome to comment and suggest freely on these, or to prove me wrong (I would be delighted !). These are just my "annoyances" that I would like to discuss about. Currently, sourceforge has a few problems. They are caused by the fact that the OSTG aims at providing service for very large and complex projects (such as gaim?), and that they serve millions users, resulting in quite a bulky swiss-knife interface, on top of some stability and speed issues. So here goes (and sorry if I may sound a little rough): - sometimes, the website and/or the project pages are REALLY really slow. - I somewhat question the stability of the services. I think that, to some extent, google might handle load better with their huge beowulf cluster of atomic supermen. - the advertising everywhere is %&%& pissing me off (whatsmore, advertising for "Get The Facts" on top of that!?) - the user interface is.. let's say, hardcore/advanced/bloated/scary? I personally do not feel so much features are needed, but I expect google will add lots of them someday. - the sourceforge bug trackers and forums should be KILLED WITH FIRE! - no matter what I do, I have to read a few pages of documentation to do something on SF. I'm lazy. - their shell and cvs service is slow. And I mean, slow like sinking deep slow with nautilus begging me to never upload new content on the website with it again! Now, of course, that would mean we would directly migrate to a google subversion repository instead of our current sourceforge CVS. That can be considered as a great plus, or it could annoy others, so I would like to hear any opinions on that. There are some things that really please me with google such as: - fast, clean, responsive interface - their issue tracker makes sense Now, here are some disadvantages I am listing, hopefully you can give me your feedback on them: - missing some good features such as website space. I guess we could keep the sourceforge-based website, though it is slow? Some other features will be added later too (such as a release area for tarballs) - everybody would need to use a google (gmail) account. Would this cause ethical problems to anyone? If not, I can also provide gmail accounts for anyone who wants one. I have over a hundred invitations waiting to be given ;) - what if google becomes evil? - should we migrate data from our launchpad.net account? I am awaiting for your comments. Flame ahead! ;-) Kiddo / Jeff |
|
From: Kiddo <nek...@gm...> - 2006-03-07 22:16:45
|
Hello, I have thought of a possible problem: specto is based on its trayicon module. What happen if the trayicon cannot be created? If the user has no tray? Specto would not make sense. Then, we must allow trayless operation for specto. When a tray icon cannot be created, the notifier window should automatically be created, and on its destroy event, destroy specto. To do so, I think we have to reconsider the current structure. We would nee= d to move lots of stuff from trayicon to a new main.py What do you think? Maybe I'm totally wrong on this one Jeff |
|
From: wout c. <wou...@sc...> - 2006-03-07 21:36:19
|
i like it very much, but the only thing i don't like, are the button-labels. i dont like the relief when you move your mouse over them. maybe you could use a column for that (with a grey background?) so the column-label can be selected. just a thought... also: i started working on some other watches (almost have imap en pop3 working! :-D ) so how are we going to handle all the different options a watch has? maybe do someting like a wizard? (first choose the watch type and name etc and then click on the next button, depending on what type you chose there will be different options available) Wout |
|
From: Conor C. <maj...@gm...> - 2006-03-07 20:32:59
|
Now it's on the mailing list :D but no attachment. OK, I like it! It looks *exactly* like the original mock-ups, which is awesome. majikstreet On Mon, 2006-03-06 at 22:20 -0500, Kiddo wrote: > I fiddled with glade tonight, trying to figure out how we could > present the updates in a nice and efficient way in the specto notifier > window. Please see the attached .glade file. > > There are a gazillion different ways we could do this (tree view > a-la-gaim-buddy-list, classic columns view, beagle-like view, a mix of > it all, or the current mockup I attached). I decided to use tables > with buttons, labels and images. There are a few reasons: > - I think a table should be pretty workable. Maybe easier than a > treeView widget, but I'm not sure. > - the interface feels cleaner, but that's very subjective > - we do not "truly" need a columns view, because there is a menu: view > sorting > > There are a few problems: > - text alignment > - left click & right click issues? possible actions: select watch, > open watch target, open a menu > > Now, I understand that I might be very very wrong on it all. And if > you have some creative ideas for the user interface, I'd be very > interested to hear them :) any comments/suggestions on the current > mockup would be appreciated too. It's in no way a final gui. > > Note: this mail was not sent through the mailing list because of the attachment. > Jeff -- majikstreet get firefox! www.getfirefox.com |
|
From: Conor C. <maj...@gm...> - 2006-03-06 23:43:07
|
[I accidentally emailed this to kiddo... doy!] Yay! mailing lists are back up :D OK, We also have a team at launchpad: https://launchpad.net/people/specto If you want to develop specto, or are a developer, please sign up on launchpad and join the team! Jeff/Kiddo will approve members as they come in. Members of the Specto Development team have access to the admin stuff for the specto product page on launchpad.. The team is listed as the maintainer of the product. Bugs can also be filed there. In addition, I was thinking that we could have Bug #1 be like Bug #1 in ubuntu: https://launchpad.net/distros/ubuntu/+bug/1 majikstreet/Conor (Lol: Kiddo kind of helped me learn a little bit of pygtk, and I made my first application.. Now I should understand the specto code more, and maybe be able to help develop the GUI later... Kiddo was patient with all my stupid questions and errors.. Thanks Kiddo :D) [added now] Also, I think as far as filing bugs go, maybe we could create separate releases/distros(not really) for specto in the bug filing thing, so that you can file a bug for the CVS specto, and the releases.. I think this is possibly but I haven't investigated! On Sat, 2006-03-04 at 15:22 -0500, Kiddo wrote: > Hello folks, > > I am suggesting using Canonical's Launchpad (www.launchpad.net, it's > the thing that replaced ubuntu's bugzilla) for a lot of Specto's > organizational tasks. I know bugzilla is more flexible than malone for > bugs, but I believe we might not be into bugzilla.gnome.org just yet. > So meanwhile, I think it could be an interesting idea to host our bugs > on https://launchpad.net/products/specto/+bugs > > I see a few advantages: > - well, it IS simple to use. > - may be not as flexible as bugzilla, but it certainly is WAY more > usable than SourceForge's bug tracker. > - you don't have to type in your friggin' sourceforge login name and > password every day. > > I then suggest also hosting our Specifications (including the roadmap) > in launchpad, in https://launchpad.net/products/specto/+specplan, and > using Rosetta to have people translate our application easily into > their native language. > > All in all, we would host > * our bugs (malone) > * specifications > * translations (rosetta) on launchpad. > > This would allow easier interaction with the ubuntu community also > (since a few of us are coming from that). Overall, let's say that I am > a bit annoyed by SourceForge's user interface and responsiveness. If > you guys agree on this, I would then close the trackers on sourceforge > (don't worry, the CVS and SVN would still be there at least), to avoid > users being lured into a non-maintainted section. > > Any comments on the issue are welcome, I like feedback :) if you see > other alternatives, I'm also interesting in hearing about them. > Jeff > -- majikstreet get firefox! www.getfirefox.com |
|
From: Kiddo <nek...@gm...> - 2006-03-04 20:48:45
|
Hello folks, I am suggesting using Canonical's Launchpad (www.launchpad.net, it's the thing that replaced ubuntu's bugzilla) for a lot of Specto's organizational tasks. I know bugzilla is more flexible than malone for bugs, but I believe we might not be into bugzilla.gnome.org just yet. So meanwhile, I think it could be an interesting idea to host our bugs on https://launchpad.net/products/specto/+bugs I see a few advantages: - well, it IS simple to use. - may be not as flexible as bugzilla, but it certainly is WAY more usable than SourceForge's bug tracker. - you don't have to type in your friggin' sourceforge login name and password every day. I then suggest also hosting our Specifications (including the roadmap) in launchpad, in https://launchpad.net/products/specto/+specplan, and using Rosetta to have people translate our application easily into their native language. All in all, we would host * our bugs (malone) * specifications * translations (rosetta) on launchpad. This would allow easier interaction with the ubuntu community also (since a few of us are coming from that). Overall, let's say that I am a bit annoyed by SourceForge's user interface and responsiveness. If you guys agree on this, I would then close the trackers on sourceforge (don't worry, the CVS and SVN would still be there at least), to avoid users being lured into a non-maintainted section. Any comments on the issue are welcome, I like feedback :) if you see other alternatives, I'm also interesting in hearing about them. Jeff |
|
From: Pascal P. <pas...@mo...> - 2006-03-04 14:13:26
|
Hello guys, Wout, in a recent commit, asked why we're using gobject.timeout_add instead of the timer module. The reason is simple : At the time that we connected the gui to the backend, to be able to add a watch directly in the add_watch window, I figured out that threads of the timer module comes in conflict with the threads of gtk. When gtk was running, no watches could update. I looked over the net and found this solution. gobject is part of gtk. When we use timeout_add, it adds the thread in the gtk threads, which makes everything run smoothly. We can even use Ctrl-C in the console now! We should use this method for any kind of timer from now on. Cheers, PP |
|
From: Pascal P. <pas...@mo...> - 2006-03-04 14:12:48
|
Hello wout,
> so what config file reader/writer are we going to use?
IMO, xml is the best way to implement this easily. The only thing that
needs to be looked at, is which type of variables we are going to use
in the program prior to converting it to xml.
The idea is pretty simple :
def __init__(self, blabla):
...
self.xml_var =3D method_to_get_info_from_conf_file()
method_to_parse_watches_written_in_conf_file(self.xml_var)
...
def add_watch(self):
...
self.xml_var.add(watch_name, type, typearg, refresh, updated)
method_to_write_to_conf_file(self.xml_var)
...
> i also changed some things in the watch class.
> when you init() the web_watch it is not started. i think this is better
> so you can still change settings (ex: interval) for the watch and then
> call start_watch() to start the watch.
> if we call it in the notifier class, it will also be much easier to show
> a message in the notifier window.
I thought about that, and I think that the real utility to this, is to
be able to deactivate some watches if we want. Because, if you think
of it, the rest of the settings can be easily set in the init.
ex :
__init__(self, bla_bla, interval =3D 3600):
self.refresh =3D interval
self.bla_bla =3D bla_bla
> i also moved the notify function to the superclass because every time a
> message has to be shown, the result is the same, only a different text
> is used, so it should be very dynamically and know what type, name,
> ...it has so it can generate a special message for the watch.
It is good that the notify method is put in the superclass, but I
think the notification message should be located in the subclass. Like
(it's really an example, not exactly what should be written):
self.notify("The web site %s has changed" % self.url)
> kiddo: nice job changing the color from the trayicon but i think it
> should come in notifier.py because it is more a gui thing than a watch
> thing.
Yeah kiddo good job! =3DP
> i hope everything is clear
After looking at the commit, it is.
Also, I would like to know, how come you take out all the specto references=
?
I think they're handy for sending back info to the trayicon instance,
which command the main functions.
Cheers,
PP
|
|
From: wout c. <wou...@sc...> - 2006-03-04 13:17:32
|
Hello, the watch class is starting to work nicely, so we should start thinking about the implementation. for now, the watch is created (and started) when you add it, but i think that the add_watch should write the watch-setting to a config file and the notifier should read the config and start the watch. so what config file reader/writer are we going to use? i also changed some things in the watch class. when you init() the web_watch it is not started. i think this is better so you can still change settings (ex: interval) for the watch and then call start_watch() to start the watch. if we call it in the notifier class, it will also be much easier to show a message in the notifier window. i also moved the notify function to the superclass because every time a message has to be shown, the result is the same, only a different text is used, so it should be very dynamically and know what type, name, ...it has so it can generate a special message for the watch. kiddo: nice job changing the color from the trayicon but i think it should come in notifier.py because it is more a gui thing than a watch thing. i hope everything is clear Wout |
|
From: wout c. <wou...@sc...> - 2006-02-08 22:08:09
|
off course it would be more logical,
but feel free to start working on it! somebody has to make the first
steps :)
when i try to use the class i get this error:
>>from specto.watch import watch
>>w = watch
>>w.set_name("test")
Traceback (most recent call last):
File "<stdin>", line 1, in ?
TypeError: unbound method set_name() must be called with watch instance
as first
argument (got str instance instead)
so there is still something wrong (or i am making a stupid mistake)
the only reason why i am working on the gui before the backend is to
learn to program in python
but i am starting to get a hold of it, so i am ready to start on the
backend.
(but don't be suprised when i make very stupid mistakes)
Wout
Conor Callahan wrote:
>Hi,
>I had an idea:
>
>Maybe we could work on the backend of the program first, then integrate
>it with the GUI.. It just makes more sense to me..
>
>Conor
>
>
|
|
From: Kiddo <nek...@gm...> - 2006-02-08 22:02:56
|
I understand your point, and that was somewhat what I was thinking at the beginning. However, my skills are far more oriented towards user interfacing. I can code, but code very simple things, so, if someone assigns me something simple I can do it, otherwise I might just end up borking the code. So I do what I "excel at" to help the project: I do more the UI stuff, while you guys can work on the backend as much as you want, I would be very happy of that. Just think of "kiddo works on the user interface in parallel to the backend, because he lacks the insight to do the backend himself at the moment". However, as Mr. Hartwell seemed pretty interested in joining the project, we might get a "backend code boost" :) If you are willing to code something on the backend, well I should not speak stupid things, so please email Pascal Potvin... he would know what could be done to further advance things.. I hope this mail reassured you a little :) I know that we have a "non functional gui" right now... Jeff On 2/8/06, Conor Callahan <maj...@gm...> wrote: > Hi, > I had an idea: > > Maybe we could work on the backend of the program first, then integrate > it with the GUI.. It just makes more sense to me.. > > Conor > -- > majikstreet > get firefox! www.getfirefox.com > > > > ------------------------------------------------------- > This SF.net email is sponsored by: Splunk Inc. Do you grep through log fi= les > for problems? Stop! Download the new AJAX search engine that makes > searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! > http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D103432&bid=3D230486&dat= =3D121642 > _______________________________________________ > Specto-devel mailing list > Spe...@li... > https://lists.sourceforge.net/lists/listinfo/specto-devel > |
|
From: Conor C. <maj...@gm...> - 2006-02-08 21:32:44
|
Hi, I had an idea: Maybe we could work on the backend of the program first, then integrate it with the GUI.. It just makes more sense to me.. Conor -- majikstreet get firefox! www.getfirefox.com |
|
From: Kiddo <nek...@gm...> - 2006-02-08 18:01:23
|
So far, I agree on the main class requirements (name + refresh
interval), but I have not yet thought of something else that is shared
among all watches, maybe a URI, but not really all watches use that
(if we create watches for package management for example).
About the pseudocode: yes, that is basically what is supposed to
happen. I think the watch_raw and watch_rss modules already use this
kind of thing, but in a different way. The watch_rss module is older,
so it needs to be revised I think...
The big big problem is that modules do not communicate (afaik)
currently. That's the blocker.
> some questions about the code that is allready in watch.py:
> def __init__(self, refresh=3D float(5.0)):
> self.refresh_ =3D refresh
> self.timer =3D Timer(float(0.1), self.update)
> self.timer.start()
> what is "self.update"? is it the interval between updates for a watch? a=
nd
> why would you use refresh =3D float(5.0), isn't that something the user h=
as to
> set?
Yes, it is supposed to be a variable. HOWEVER, we set it to something
like 5 seconds, but that's ONLY for development and testing purposes.
It should be removed when watches can "talk".
> And when i wan't to create an object from the module i get this error:
> >>> import specto.watch
> >>> t =3D specto.watch()
> Traceback (most recent call last):
> File "<stdin>", line 1, in ?
> TypeError: 'module' object is not callable
> >>>
I'm not sure, wasn't there a watch call example lying there somewhere?
hmmm... Try this, but I can absolutely not guarantee it will work,
it's just an old testwatch.py lying in my archives...
#!/usr/bin/env python
from watch_raw import rawCacheCreator
def hasChanged(text):
print text
raw =3D rawCacheCreator("http://nanokron.info", hasChanged, "",
['Nouveau Fichier 1'])
raw2 =3D rawCacheCreator("http://127.0.0.1/Screenshot.png", hasChanged,
"", ['Nouveau Fichier 2'])
raw3 =3D rawCacheCreator("http://specto.sf.net", hasChanged, "",
['Nouveau Fichier 3'])
I think this calls watch_raw.py and runs three different "watch"
instances. Of course you have to change the URIs to whatever test
files you want to use and modify and see if it works.
Now, some terminology so that I'm sure you are not confused at the way
I named things in the beginning:
watch_rss =3D a watch module for RSS/atom feeds
watch_raw =3D caches, MD5sums and compares ANYTHING. That's for static
webpages, images, whatever. Pascal coded this based on something I
don't quite remember the name at the moment.
self.update =3D is a function that CHECKS FOR UPDATES, then if there are
updates, calls self.notify
self.notify =3D is where we should send the signals to the notify module
that the watch was updated, so that the gui displays the relevant
information. However, currently (can't say that enough ;), the modules
don't really communicate, so this only prints that the watch was
updated, in the terminal.
I hope I could answer some of your concerns. If I'm wrong on any of
these, Pascal please correct me.
Jeff
|
|
From: wout c. <wou...@sc...> - 2006-02-07 19:27:48
|
It's time that we start thinking about the main functionalities from the
watch class
so we can actually start working on it.
a main watch has a:
-unique name
-interval timer / watch (time between updates)
-...?
the specific options for a type of watch are set in the subclass.
the main watch class has a function for:
-returning a message when a watch is updated
-setting/getting a name for a watch
-setting/getting the interval period
-calling the subclass when the watch has to be checked for updates
-...?
so it would look something like this (in pseudo):
/if timer == interval
call subclass
(subclass
check_update
if updated:
return updated = True
else:
return updated = False
end subclass)
if updated == True:
return "Watch updated"/
This doesn't look difficult, does it?
some questions about the code that is allready in watch.py:
/ def __init__(self, refresh= float(5.0)):
self.refresh_ = refresh
self.timer = Timer(float(0.1), self.update)
self.timer.start()/
what is "self.update"? is it the interval between updates for a watch?
and why would you use refresh = float(5.0), isn't that something the
user has to set?
And when i wan't to create an object from the module i get this error:
/>>> import specto.watch
>>> t = specto.watch()
Traceback (most recent call last):
File "<stdin>", line 1, in ?
TypeError: 'module' object is not callable
>>>
/How can we solve this?
these are my thoughts about the main watch class, feel free to change it!
Wout
|
