Kar(In reply to comment #21)
> Timothy, thanks for the reply. I am of the opinion that getting this patch in
> for FF3 is very important for tagging and places intergration plus useability.
> People really are not going to tag 1000+ bookmarks one by one. Do you think you
> will be able to get a patch working for final?
> 

Yes, I understand the concerns in regards to usability and I think this would
provide value to the great improvements of Places.

I believe I can deliver a working patch for final. I am planning to work on it
again this weekend. I am just busy with doing away with exams and all right
now. By next Sat I should be confident to deliver.

DPS911/909 OSD/600/900 – You never know when and who values your hard work!

Keep up the motivation to deliver amazing work!

- ctrl+a within the tag folder, causes unexpected results: the bookmarks looses their tags.
- standard multiple select key stokes — ctrl+click and ctrl+shift + click comibnations, also causes unexpected results.
- the key stroke, ctrl+a works with regular folders (essentially, non-tags subfolders).

STR:
- tag a set of bookmarks with the value of “a”.
- click the “a” tag subfolder.
- enter “, b” to include a new tag.
- select the “search bookmarks” text box to update the tags (ie. trigger the onblur event)

Results: The bookmarks within the “a” tag subfolder disappears. The bookmark location within the regular folders remain the same.

System enhancements are difficult to perform testing especially if there is no established unit testing framework. It can be very challenging to manage all impacts made especially if you are unfamiliar with the given code base. But nonetheless, reading and understanding all these impacts is essential in order to properly take the necessary actions.

It’s really a privilege to be in school and have the opportunity to refine your skills and knowledge within a luxurious set of deadlines. For the most part, in the real world deadlines really mean ASAP…Well thats all for now. I’m going to keep you guys posted on my self-learning (or re-learning) and hopefully share more of my experience and lessons learned.

Multi Edit or Bug 412002 -should be able to edit tags for multiple bookmarks at the same time, is the current active bug that I am working on. The official bug title explains what the bug hopes to achieve.

Figure 1 – Firefox 3: Places Bookmarks
The new Firefox 3 contains a significant amount of improvements and changes. One change that I want to highlight is the edit bookmark panel. In the FF3 source it is called editBookmarksOverlay.xul. The above link is showing my bookmarks items in the Bookmarks Menu folder.

Figure 2 – Bookmark Item Selected
When a bookmark item is selected the edit bookmark overlay panel loads text boxes that allow for bookmark attributes to be updated.

Figure 3 – Multi-Bookmarks Selected
When a user selects more than one bookmark item, either by using a ctrl+[mouse click] or shift+[mouse click]. The bookmark edit overlay panel loads only the ability to update the Tags bookmark attribute.

Figure 4 – Updated Tags
Here you can see that I have updated the tags from the bookmark items selected in Figure 4. I included the tag “sports” for the NBA.com and Raptors.com bookmark item. One of the ways the update tag is triggered, is when anytime the textbox focus of the Tags bookmark attribute field is removed. The textbox focus refers to the event when the cursor is placed and exists within a given textbox.

There are problems with a conflicting event. An event triggers a function to change the edit bookmark panel every time a bookmark item is selected. In my current release I have provided a hack that seeks to ensure that the update tag function is called before the algorithm changes the edit bookmark panel. This hack requires more testing as I have noticed inconsistent results.

Figure 5 – Related Problems 1
The first related problem that I have identified with my patch is with the Bookmarks pop-up editor which is triggered every time a Ctrl + k is issued within the browser. The editor is suppose to be more compressed and the “Folder:” field should contain a bookmarks folder location, like Bookmarks Menu or Unfiled Bookmarks.

Figure 6 – Related Problem 2
Continued from the previous Figure, when a user selects the drop down button from the “Folder:” field an Assertion Failure is issued.

First and foremost this was my first attendance of an in-class Mozilla guest speaker session. During DPS909 – Topics in Open Source Development, I was working full-time at RBC GTO for my extended co-op term so therefore I was unable to attend the all star line up. I still find it amazing and a great privilege that Mozilla / Firefox developers and full time workers would come in and interact with students. This is a very encouraging experience and makes me proud of what Seneca and their staff are doing for their computer science program.

During the start of the session, we had a round table discussion where we had an opportunity to introduce ourselves and to talk about our current project. It was very fascinating to actually put the IRC names / projects to the student faces. After the round table discussion we had the opportunity to ask Shaver questions.

The most interesting question was how Mozilla and other open source model their finance. One of their financing opportunities comes from Google, where the a contractual relationship is based on the integrated web search tool (when you press control+k, the cursor will focuses on it). Despite other details mentioned, the striking thing that came out was that the deal was motivated from a growing amount of Firefox users, using Google as their main search engine. I found it very impressive that Mozilla captured business opportunities while trying to keep their services and products community centric.

The community centric theme, later returned from a discussion related to forking. Forking is a term used to describe software that is based on another piece of software – in terms of source code and package but has been modified to be distinct. Examples are Iceweasel and Flock forking Firefox.  During the discussion, David Humphrey shared his open source project experience. Humphrey and Chris Tyler (I believe he is also working on this Project or is at least a major contributor) are attempting to package a Mozilla Development Kit on DVD for future students. This kit comes with everything a student needs to do Mozilla Development. Humphrey informed us about his failed attempts to get help and support from the Visual Studio Team so that their tools could be included into the kit. Essentially, the problem with the Visual Studio Team was over licensing concerns and other related issues.

The point that I got out if it was that with the open source model, the path for innovation is much easier which means the community has the advantage. This reminded me of Bob Young’s discussion over this years FSOSS and how the open source model benefits products and services that reaches out to a large community or that it contains a community centric theme…

So right now I am hoping and praying that the Open Office community becomes bigger and better so that their product line catches up to and increase in quality over MS Office because right now the MS Office market and software is kicking butt!

Here are the ways I think we should consider exposing the ability to tag

multiple pages at the same time:Adding tags:

-multiple selection, editing the tag field in the properties pane
-multiple selection, drag operation to the tag in the left pane
-multiple selection, copy and paste to a tag folder

The one that is fascinating and challenging about developing bug fixes is that you are forced not to program for yourself which again is one key lesson learned in Barb’s Systems classes -my own personal opinion. The job is to take something that has been created and updated by numerous people, use all that and creatively attempt to apply your solution making sure that it not only works but its cohesive with the entire application / system… that’s why I agree with my Stock Broker neighbor’s perspective on computer science: “Computer science? I thought that’s more of a an art than a science…”

Lesson Learned

A lesson learned based on my previous patch thats worth sharing to new mozilla code contributors is when submitting prototype patches –that is code thats not ready to be reviewed for checkin into the source, Firefox full time engineers prefer the following:

> So basically you want me to clean up the code? So in the future when I
> submit prototype code I will just attach the file but not assign it for
> review? The assign for review is something more formal? That's what I wanted
> to talk to you about.

here's how i do it:

1. upload "WIP" (work in progress) patches, no review request, to save
my work, and for drive-by reviews of the approach
2. once you think a patch is ready to be added to the product, ask for review

Progress notes on release 0.6

My initial search for a solution to release 0.6 and some results that stood out and I’m going to further investigate. I just want to describe and share what and how I searched for a solution within MXR.

Used “drag and drop” as a value for “Text Search” field.

/browser/base/content/browser-places.js,

* line 740 — * Drag and Drop handling specifically for the Bookmarks Menu item in the

/browser/components/places/content/controller.js,

* line 1508 — * Handles drag and drop operations for views. Note that this is view agnostic!

Used “drag and drop” as a value for “Search for” field and filtered the “in files matching” field to query anything with a value of “places”. The filtering by file name (“in files match”) happens when you initially perform a “Text Search” in http://mxr.mozilla.org/firefox/.

/browser/components/places/content/controller.js,

* line 115 — case “cmd_paste”:
* line 116 — return this._canInsert() && this._isClipboardDataPasteable();
* line 189 — case “cmd_paste”:
* line 215 — case “cmd_paste”:
* line 216 — this.paste();
* line 342 — * Looks at the data on the clipboard to see if it is paste-able.
* line 343 — * Paste-able data is:
* line 349 — _isClipboardDataPasteable: function PC__isClipboardDataPasteable() {
* line 351 — // pasteable, with no need to unwrap all the nodes.
* line 1138 — * Paste Bookmarks and Folders from the clipboard
* line 1140 — paste: function PC_paste() {
* line 1171 — * Gets a list of transactions to perform the paste of specific types.
* line 1173 — * The types of data to form paste transactions for
* line 1174 — * @returns An array of transactions that perform the paste.
* line 1208 — // Get transactions to paste any folders, separators or links that might
* line 1213 — var txn = PlacesUtils.ptm.aggregateTransactions(“Paste”, transactions);
* line 1216 — // select the pasted items, they should be consecutive

/browser/components/places/content/utils.js,

* line 825 — * Constructs a Transaction for the drop or paste of a blob of data into
* line 828 — * The unwrapped data blob of dropped or pasted data.
* line 832 — * The container the data was dropped or pasted into

Release 0.5 Reflections Notes

The process of finding where I need to code came from a top-down approach where I started reading the UI layer to find the targeted functions/components (item selection/properties panel view states and the tag persistence layer). The techniques I used are mainly educated guesses/assumptions and debug “alert” statements to help me trace and walkthrough the code. In this release I also leveraged the MXR repository tool to assist in understanding code. For instance there was a discrepancies I found with a method in the Places View so I shifted my focused to MXR for an answer. The last notable highlight of release 0.5 and the code review is the persistence layer. The persistence layer is encapsulated in a XPCOM class called nsITaggingService. I created a mind map but its a hard copy. I still need to get use to creating a soft copy as I like the traditional manual way –the usability and user experience is much better!

Release 0.5 Status

There are three problems I had encountered. Two are bug related while the other is more on process. The bug related problems can be found as a posted comment filed in Bug 412002. I have also attached my latest patch there as well.
My process problem was a possible patch collision. A patch collision problem is when you create a patch from an outdated source tree on your local directory and attempt to apply it on the current source tree in the repository. I consulted my classmate Cesar about this issue. He informed me that testing out my outdated patch on a separate folder containing the updated source tree would be the best idea. Luckily I had only one block that failed during the patch and was small enough to manually fix it.

Now I want to find out a more efficient method in resolving collision problems with patches!

1. Find out the function that displays the properties form when (i) a bookmark item is selected and (ii) when a bookmark folder item is selected.
2. Find out the function that takes care of persistence management.
3. Modify the code to include the properties form to be enabled when more than one item is selected. Then update the way the persistence management layer so that it includes the latter operation.
4. I am not sure how identifying two or more items selected works, so I am going to need some additional research in this field if upon finishing item 1 of this list provides no certain solution.

During the discussion I found on that the tag editor bug has now gone into limbo. It seems that the tag editor was going to implemented by an external party so Dietrich suggested that its not a good bug to dwell on at the moment. I told him that I was willing to have any replacement for that bug (So it looks like going to have some time available to do more ‘for each’ js fixes on the trunk for mfinkle after all… )

So for my plans for release 05, we both agreed that working on the multi-ended bug would be the best focus. What I’m going to do is to first prototype the ability to “select multiple URIs and edit tags”.

Mind Mapping and Code Understanding

One method of researching and understanding code I want to commit to do is to try to produce a mind map. Mind mapping was introduce to me in my 12th grade physics class as a form of brainstorm and test/exam preparation technique. The motivation behind this was during last years work on the Coop. I had difficulties in trying to remember what function called each other and thus my editor was scattered with bookmarks.

My goal for mind mapping is not to produce every occurrences of each function as that is already provided in MXR but rather assist in providing a visual conceptual model view of what functions and components relate with each other within the Places API and Firefox itself.

I’m a visual learner so I hope the extra work in producing this can benefit me in the long run. I will produce this mind map on my main wiki page to share.

I found an open source mind mapping software called FreeMind. It is in its 0.x release so expect some rough edges but overall it does the job. It can export the mind map into various file versions like a doc or HTML.

Action Items Planned

- Find out where the changes need to be applied
- Research proposed solutions (MXR, Google, IRC)
- Try and test out solutions

My initial research was sifting through documentation on the Mozilla website. I found articles, documents on the goals and requirements for the Places API and was able to find the main channel of the Places developers on IRC. Before obtaining the bugs list, I wanted to try to connect with the Places developers and just let them know that I was interested in helping out the team by providing bug fixes. I want to be able to build a relationship with this community so that I can know how to work with them as a team and meet both my objectives and theirs.

I had a very good conversation in IRC with Dietrich (eventually I learned Dietrich was going to be my mentor), a Firefox engineer working on Places. Dietrich gave me an overview of the Places code location within the trunk:

<dietrich>
Places is split (as you can see on the wiki) between /browser and /toolkit
<dietrich>
places code in /browser is mostly xul and js
<dietrich>
places code in /toolkit/components/places is mostly xpcom services
<dietrich>
some in js, some in c++

He also gave me tips on joining the Places mailing list and watching others handle and attack bugs. Other cool Moz folks helped me to get edit bug status on my Mozilla bugzilla account (Thanks Gijs, mfinkle and gavin!).

Dietrich initially suggested that I work on the multi-ended bug as the tag editor bug had some dependencies that needed to get resolved but we then decided that working on Bug 412600 would be more fitting for me to tackle first. This way it would allow me to do some code scanning of the Places code base and help me to get accustomed to the bugzilla environment. Although the bugzilla interface is quite intuitive and the topic was introduced in last years course, I still wanted to find references to help guide me when submitting my FIRST patch. Here are some articles I stumbled over the Internet that I want to share before ending this post.

Enjoy!
http://www.bugzilla.org/docs/2.18/html/lifecycle.html

http://developer.mozilla.org/en/docs/Creating_a_patch

http://developer.mozilla.org/en/docs/Getting_your_patch_in_the_tree

http://fdiv.net/2007/05/15/the-secret-life-of-a-thunderbird-patch/#more-74